TSMC on Thursday announced its financial results for the second quarter of 2024, reporting a 40.1% year-on-year revenue increase. The Taiwan-based chip foundry also stated that it has raised its full-year revenue and capital expenditure targets for 2024. The firm’s earnings call was also remarkable for TSMC CEO C.C. Wei introducing the new concept of “Wafer Foundry 2.0” for the first time during his comments to investors.

Why it matters: TSMC’s introduction of Wafer Foundry 2.0 signifies a strategic expansion beyond traditional wafer manufacturing. The move aims to capitalize on emerging market opportunities and meet the growing complexity of semiconductor production.

Details: TSMC said the idea of Wafer Foundry 2.0 is intended to better leverage its advanced packaging capabilities for market expansion by encompassing a wider range of activities compared to the traditional wafer manufacturing industry.

• As chip manufacturing becomes increasingly complex, wafer foundries have evolved beyond simple wafer manufacturing services. Wafer Foundry 2.0 was unveiled during TSMC’s second quarter earnings call, during which Wei redefined the foundry industry to include packaging, testing, mask-making, and IDM (Integrated Device Manufacturer) operations.
• Wei also noted that demand for TSMC’s 3nm and 5nm process technologies remains strong. As the company enters the third quarter, the demand for advanced processes driven by smartphones and AI opportunities has slightly raised its full-year revenue forecast for 2024, increasing the expected growth range from 21%-26% to 24%-26% year-on-year, he said.
• TSMC has also raised its 2024 capital expenditure budget from $28 billion-$32 billion to $30 billion-$32 billion. Of this, 70% to 80% will be allocated to advanced process technologies, 10% to 20% to special process technologies, and 10% to advanced packaging, testing, and mask-making, the company said.
• In the second quarter, TSMC reported revenue of NT$673.51 billion ($20.56 billion), a 40.1% increase year-on-year and a 13.6% rise quarter-on-quarter. Net profit was NT$247.85 billion ($7.57 billion), up 36.3% from the previous year and 9.9% from the previous quarter.
• In terms of revenue distribution by process technology, TSMC’s revenue from advanced processes of 7nm and below reached 67% in the second quarter. The firm’s 3nm process revenue accounted for 15% of this total, largely due to the launch of Apple’s A17 Pro series processors last year, as well as its M4 series processors and the upcoming A18 series processors this year.
• By end-use application, the revenue breakdown is as follows: high-performance computing accounted for 52%, smartphones for 33%, IoT for 6%, automotive electronics for 5%, consumer electronics for 2%, and others for 2%.

Context: Apple has secured the first batch of TSMC’s 2nm production capacity, expected to start trial production soon. The new chips, debuting in the iPhone 17 series, may boost performance by up to 15% and cut power consumption by 30% compared with the 3nm process.

Kuaishou, one of the main rivals to TikTok’s China sibling Douyin, showcased several fresh features for its text-to-video model Kling AI at the World Artificial Intelligence Conference (WAIC) in Shanghai last week, including the ability to generate videos up to 10 seconds. 

At WAIC, visitors queued to experience the Sora-like tool that is currently available by invitation only. Users sent simple prompts to generate videos, such as “a panda eating salmon” and “the Mona Lisa putting her glasses on with her hands”, with the resulting clips demonstrating Kling AI’s ability to render the inputs almost perfectly.

AI-generated videos have subsequently flooded the Chinese internet, with Kling AI being used to create clips featuring characters from historical films undertaking modern day tasks and spawning multiple memes.  

A video featuring Rong momo, a character from “My Fair Princess” who has become a well-known internet meme in China, feeding Princess Ziwei a chicken drumstick has gone viral on social platforms these days. The AI-generated video is based on the drama’s most famous scene in which Rong momo tortures Ziwei by repeatedly stabbing her with a needle.

Why it matters: Kuaishou will be hoping that its suite of self-developed large model series, including language model KwaiYii, image-focused Kolors, and video-centered Kling, will give it an edge as it continues to challenge to ByteDance’s Douyin and TikTok.

Details: More than 500,000 users have applied to help beta test Kling, senior vice president of Kuaishou Gai Kun revealed last weekend at a WAIC forum, with the number of videos generated reaching 7 million as of now. The Sora rival’s hype is such that English language posts teaching users outside of China how to apply for a Kling AI trial can be found on X, formerly known as Twitter.

• Kuaishou provided practical tips on the screen at the WAIC event, including advising users to use simple words and sentence structures and to avoid overly complex language. It also emphasized that its model is not sensitive to numbers, giving an example that if the prompt is “10 puppies on the beach,” the number might not be consistently maintained in its outputs.
• A member of staff from Kuaishou’s large language model team told TechNode that they were not at liberty to disclose the data used to train Kling AI, but indicated that it was open source.
• The TikTok rival meanwhile announced at WAIC that its Midjourney-like model Kolors would become open source, a move which Kuaishou said aims to contribute to a more prosperous ecosystem for the text-to-image generation community.
• Kuaishou’s investment in research and development has quadrupled over four years, with expenditure increasing from RMB 2.9 billion in 2019 to RMB 12.3 billion in 2023.

Context: Kuaishou, China’s second-largest short video company, launched its AI strategy in 2023, according to CEO Cheng Yixiao, who said that generative AI has a “very rich combination of business scenarios and huge value potential” for the content platform.

Editor’s note: ‘Landing AI’ is a series of special reports focusing on the field of Artificial Intelligence curated by TechNode. By investigating the development of AI landing in China and the behind-the-scenes stories of the industry, we’re going to dive deeper into everything that’s possible under the new wave of AI.

“Hello, I will arrange the bread and fruit separately.” Upon hearing the command from the host, the humanoid robot Qinglong extended its arms and gently picked up a piece of soft bread with its steel fingers, placing it down into the basket on the table. 

At the World Artificial Intelligence Conference (WAIC) 2024 in Shanghai, one of the most influential AI events within the global tech ecosystem, the National Local Joint Humanoid Robot Innovation Center on Thursday unveiled the open-source general-purpose humanoid robot Qinglong for the first time. The event marks the debut of China’s first full-sized general-purpose humanoid robot model, along with the release of its related open-source technologies.

Why it matters: The unveiling of Qinglong marks China’s entry into the advanced field of full-sized general-purpose humanoid robots. Its open-source initiative is intended to accelerate innovation, collaboration, and development within the global robotics and AI communities, potentially driving advancements in a range of industries.

Details: During the Summit on Humanoid Robots and Embodied Intelligence Development Forum at WAIC 2024, the National Local Joint Humanoid Robot Innovation Center demonstrated Qinglong’s operation and design concepts on site.

• Qinglong has been independently developed by Humanoid Robots (Shanghai) Limited, a new R&D institution established by leading industry enterprises with a registered capital of one billion RMB ($140 million). The company was designated as the National Local Joint Humanoid Robot Innovation Center by the Ministry of Industry and Information Technology of China in May 2024.
• Qinglong stands at 185cm tall and weighs 80kg, according to the company. It features a highly bionic torso and anthropomorphic motion control, supporting multi-modal mobility, perception, interaction, and manipulation. With 43 active degrees of freedom, Qinglong achieved a maximum joint peak torque of 400 N·m (Newton-meters) and 400 TOPs (Tera Operations Per Second) of computational power.
• The humanoid robot combines mobile lower limbs for agile walking with lightweight, high-precision upper limbs for operations, the on-site presentation stated. The company claimed that the model supports rapid walking, agile obstacle avoidance, stable incline and decline movement, and resistance to impact interference, making it an ideal platform for developing general artificial intelligence software and hardware.
• “In the Year of the Dragon this year, we named our humanoid robot Qinglong [meaning green dragon in Chinese] and initiated an open-source community for humanoid robotics,” said Jiang Lei, Chief Scientist of the National Local Joint Humanoid Robot Innovation Center. “Through open-sourcing, our goal is to introduce a new humanoid robot model annually, each named after one of the Chinese zodiac animals, to foster a humanoid robot innovation community unique to China,” he announced at the forum.
• Jiang also announced the launch of the OpenLoong open-source community website, developed by the Innovation Center. The website includes the release of the robot’s hardware structure and parameters, with the embodied intelligence software package set to be open-sourced soon.

Context: On the same day, Tesla debuted its second-generation humanoid robot Optimus at the Shanghai World Expo Exhibition Hall during WAIC 2024. According to Tesla booth staff, compared with the first generation, Optimus has improved its walking speed by 30% while maintaining upright posture. Its fingers have also evolved to perform delicate tasks such as gripping eggs and handling heavy objects.

• Tesla plans to start limited production of humanoid robots next year, with over 1,000 Optimus units expected to assist in production tasks at Tesla factories, according to CEO Elon Musk.

Chinese artificial intelligence firm iFlytek on Thursday unveiled the latest version of its AI model, Spark 4.0, which the company benchmarked against OpenAI’s GPT-4 Turbo, saying it shows better performance than the US company’s most-advanced model in five aspects, despite lagging in terms of coding and multimodal abilities.

Why it matters: Trained exclusively on a home-grown computing platform that is co-built with telecom giant Huawei, Spark 4.0 can be seen as sending a signal that China’s AI industry has entered a new phase of self-sufficiency despite the US’s restrictions on the country’s access to the most advanced chips, according to iFlytek.

Details: The Hefei-based company finally fulfilled its commitment to upgrade its Spark model to the level of GPT-4 in the first half of 2024, eight months after the promise was made. At Thursday’s event, however, Chairman Liu Qingfeng acknowledged that the gap would “widen again”, possibly to more than a year, once GPT-5 is released, because of its larger parameter size, longer training time, and the increase in the amount of data used.

• “It is necessary to recognize the comprehensive gap between China and the US in the LLM in a scientific and rational way,” said Liu, “but also to have the confidence to catch up quickly to ensure there is no generation gap.”
• The company has seen over 400,000 overseas developers build products based on the Spark V3.5 model that was introduced in January, with 7.02 million developers having used it in total.
• In a key upgrade, the Spark app and its desktop version, both of which are by iFlytek’s model series, added a feature called “personal space” where foundation models are dedicated to generating individualized content that matches one’s personality. The mechanism operates on the premise that users can input their information on work, study, and life, or accumulate sufficient dialog records with AI bots for enhanced learning.

Context: While iFlytek said its Spark 4.0 model ranked number one in eight international mainstream test sets, rival Alibaba on the same day was also certified by the world-renowned open-source platform Hugging Face as topping the world’s open-source big model list for its Qwen-2, with Meta’s AI model ranked behind it in second.

China’s AI companies are rushing to win over OpenAI users in the country as the ChatGPT creator takes measures to block API traffic from countries and regions that it does not officially support. Affected regions include mainland China, Hong Kong, Russia, and Iran.

Why it matters: OpenAI’s latest announcement on API restrictions has been widely interpreted as targeting China, where  local enterprises and developers have been using OpenAI’s API to build products or services. On the other hand, the move is being regarded as an excellent opportunity for Chinese AI companies to woo these users.

Details: Developers in China have posted screenshots of a notice from OpenAI to various online communities that states, “Our data shows that your organization has API traffic from a region that OpenAI does not currently support.” The notice adds that OpenAI will block such access starting on July 9.

• API is an abbreviation for Application Programming Interface. Although Chinese IP addresses are not allowed to use OpenAI’s artificial intelligence software directly, local developers and companies have been able to create applications through OpenAI’s API platform, while individuals can change their IP address in order to get ChatGPT services. 
• Search giant Baidu, one of China’s leading AI companies, on Tuesday launched a program called Cloud in Hometown that grants additional tokens that equal the scale of their OpenAI usage to users who migrate from OpenAI to the company’s own ERNIE series model. 
• Baidu is also touting a zero-cost solution for developers to migrate from OpenAI, although it does charge a fee for further training, fine-tuning of prompts, and custom development services.
• Alibaba Cloud has also joined the fray with an alternative solution that it claims is “the most cost-effective” with 22 million free tokens for Chinese developers.
• Local AI startups, Zhipu AI, 01.AI, and Baichuan, have also launched exclusive offers for OpenAI users, with Tsinghua University-backed Zhipu being the most generous at time of writing with an offer of 150 million tokens for free.

Context: China has over one hundred AI models with parameters beyond 1 billion, according to data disclosed by the National Bureau of Statistics in March. Commercialization has become a key priority for the companies behind these models in recent months with a price war over tokens breaking out as the firms vie for potential clients.

A Chinese AI content creator has challenged cybersecurity firm 360 Security Technology over its use of an image of a woman wearing an ancient costume, which he claims he created via an AI model to showcase a “partially redrawn feature” on the company’s search engine.

The argument between creator DynamicWang and 360 Security escalated as the two parties failed to settle, with the company’s vice president Liang Zhihui saying it is willing to resort to legal action.

Why it matters: As the emerging technology of artificial intelligence is increasingly being used to create content, authorship attribution lacks a clearly defined legal framework.

Details: Even before DynamicWang publicly asked 360 Security to apologize over the alleged copyright infringement on June 8, the firm was in the midst of a public opinion storm, this time over its perceived disrespect toward women. When founder Zhou Hongyi introduced the repainting function at the company’s AI product launch on June 6, he used “sexy” as a prompt to ask the AI-powered search engine to redraw a woman’s breasts of the female in the controversial picture, which DynamicWang since claimed was based on his AI-generated work.

• The creator said 360 Security should acknowledge the infringement and compensate him with a symbolic RMB 1 ($0.14). He said the firm “stole” the image generated by his self-trained AWPortrait model.
• According to Liang, the author made a request to the company that he wants 360 Security paid him ten times the model price, together with the cost of compensation. 
• Local media outlet NBD quoted DynamicWang as interpreting the price to use his model as RMB 4,499 is for small- and medium-sized AI startups. This was not the price set for public-listed 360 Security, the creator said. 
• Each side has its own version of the story. While the creator mentioned compensation and apology were “prerequisites,” and buying the model was optional, Liang said purchasing the author’s AWPortrait model for ten times the price was “beyond our knowledge,” according to a WeChat screenshot of his conversation with DynamicWang. The current argument between the two sides is whether DynamicWang owns the copyright to the images generated by the model he trained. 
• In his latest Weibo post late Wednesday, the creator introduced himself as a photographer whose images are posted on Unsplash, Getty Images’ free gallery site, and that the training data he used to feed AWPortait model included his photography and Unsplash images, which he says means there are “no copyright issues” on his side.

Context: In January, the Beijing Internet Court granted copyright protection for an artificial intelligence-generated image, ruling that the image involved had an element of “originality” due to the plaintiff inputting multiple prompts and adjusting the parameters before generating a picture of a young woman through the text-to-image AI model Stable Diffusion. The ruling was seen as the first such AI-generated image copyright infringement case in China.

The demand for advanced process chips has surged, as AI servers and high-performance computing (HPC) applications transition to AI phones. Taiwanese media outlet Economic Daily News reported on Tuesday that Apple, Qualcomm, Nvidia, and AMD have almost booked TSMC’s 3nm process to full capacity, leading to a queue of customers extending to 2026.

Why it matters: The rapid evolution of AI technology across multiple sectors has driven a surge in demand for advanced process chips. Increasing orders from major clients is expected to further push TSMC to innovate and develop advanced manufacturing processes. 

Details: Currently, TSMC’s 3nm lineup includes N3, N3E, N3P, N3X, and N3A, according to the Economic Daily News report.

• N3E, which entered mass production in the fourth quarter of last year, primarily serves AI accelerators, high-end smartphones, and data centers. N3P is expected to enter mass production in the latter half of this year and will become the preferred chip process for mainstream applications including mobile devices, consumer electronic products, base stations, and networking by 2026. N3X and N3A are designed for customized demands such as high-performance computing and automotive clients.
• The latest high-end chips this year are all adopting the 3nm process, according to Chinese media outlet Icsmart. Apple will introduce the M4 series processors this year and is set to release the next-gen A18 series processors for its new iPhone16 series this September. Later this year, Qualcomm will launch its flagship mobile platform Snapdragon 8 Gen 4, while Nvidia is set to unveil the RTX 50 series graphics cards. AMD will release the fifth-gen EPYC Turin processors in the second half of the year, with plans to launch the MI350 series next year.
• TSMC, as the exclusive supplier of Apple’s A-series processors for the new iPhones, will see a significant increase in demand for related 3nm orders. Apple’s iPhone 16 series this year is expected to increase sales volume by 5% year-over-year to 92-95 million units, the Economic Daily News report forecasted.
• To ensure stable supply over the next two years, TSMC’s strategy is to convert some of its 5nm equipment to support 3nm production. Industry sources indicate that TSMC is likely to up its production of 3nm wafers to between 120,000 and 180,000 a month.

Context:The total production value of the top ten semiconductor foundries in the first quarter reached $29.2 billion, a decrease of 4.3% compared to the previous quarter, according to market research firm TrendForce.

• In the first quarter, TSMC led with a revenue of $18.85 billion and a market share of 61.7%, followed by Samsung with $3.36 billion and an 11% market share. Shanghai-based chip foundry SMIC took third place with $1.75 billion in revenue and a 5.7% market share.

Chinese artificial intelligence company Zhipu AI on Wednesday announced its second price cut in a month, slashing the price tag of its “most popular model” by more than half. The company said it would now offer its GLM-4 model at a charge of RMB 0.10 ($0.014) per 1,000 tokens, but CEO Zhang Peng insisted the move was not part of “a simple price war.”

Why it matters: Offering ever lower prices to capture market share characterizes China’s AI sector at the moment, where over 200 large language models have been released since October 2023.

Details: Tokens are sequences of data that large language models use to process text. GLM-4-0520, the most advanced model built by Zhipu AI, has a context window of 128,000 tokens with RMB 1 for every 10,000 input tokens. The price of this advanced model was cut in May.

Generally, 1,000 tokens are equivalent to around 750 English words or 500 Chinese characters.

• The deep discounts are a result of “iterations of the models’ core technologies,” Zhang said at a Zhipu AI open day on Wednesday. In an interview on the same day, Zhang was quoted by local media outlet TMTPost as saying, “not only can we join [the price war], but we can do it without losing money.”
• The company, which is backed by leading Chinese academy Tsinghua University, has disclosed that it currently serves 300,000 enterprise-level customers, and on average its model family processing exceeds 40 billion tokens daily.
• On the consumer side, the firm has a GenAI app called Zhipu Qingyan that is powered by the company’s foundation model ChatGLM, with the app already having amassed 300,000 AI agents that can perform tasks such as creating mind maps and making schedules since its launch last August, the company said.

Context: Zhipu AI has received at least nine rounds of financial injections from Chinese state-backed funds, major tech companies, and well-known venture capital firms since it was founded in 2019. Those supporters include Alibaba, Tencent, Sequoia, and Hillhouse.

• The Beijing-based company also acts as an investor and has invested in seed and angel rounds in rival ModelBest, a firm similar to Zhipu AI that was founded by Tsinghua University graduates. 

Two authors behind a Stanford University AI project have apologized to the Chinese team behind open-source AI model MiniCPM-Llama3-V 2.5 after social media users in China outed the former for having plagiarized the latter model, which was developed by Tsinghua University and ModelBest Inc. The incident sparked widespread discussion on the Chinese internet.

Two authors of the Stanford Llama3-V team, Siddharth Sharma and Aksh Garg, today apologized to the MiniCPM team on social media site X for their academic misconduct, announcing they would withdraw the Llama3-V model from use.

Why it matters: Members of the Stanford team acknowledged they had plagiarized Tsinghua University and ModelBest Inc’s work. 

Details: “We sincerely apologize to the authors of MiniCPM for our failure to verify the originality of Llama3-V,” Aksh and Siddharth wrote on X today. “Mustafa, who wrote the code, described exciting extensions that we promoted without knowing about the prior work by OpenBMB (founded by Tsinghua University and ModelBest Inc). We take full responsibility for this oversight. We’ve removed all references to Llama3-V in respect to the original authors.” 

• On May 29, a Stanford AI team claimed online that they could train a multi-modal large model surpassing GPT-4V for only $500, according to local media outlet Quantum Bit. Subsequently, social media users discovered that the team’s Llama3-V model used a model structure and code highly similar to ModelBest’s MiniCPM-Llama3-V2.5, with partial variable name modifications. Llama3-V also features the same tokenizer as MiniCPM-Llama3-V 2.5, including the latter’s newly defined special symbols.
• On June 2, Chinese firm ModelBest Inc confirmed that the Stanford large model project Llama3-V, similar to MiniCPM, is able to identify Qinghua Jian ancient Chinese characters from China’s Warring States Period. Remarkably, the matches also shared identical mistakes. This character data, obtained by the research team through months of scanning and manual annotation of each Qinghua Jian character, had not been publicly disclosed, confirming the act of plagiarism.
• “While it’s good to be recognized by international teams, we believe in building a community that’s open, cooperative, and trustworthy,” ModelBest Inc CEO Li Dahai stated. “We want our team’s work to be noticed and respected, but not in this manner.”
• ModelBest Inc’s chief scientist, Liu Zhiyuan, who is also a tenured associate professor at Tsinghua University, told an Yicai reporter that the rapid development of artificial intelligence relies on the global sharing of algorithms, data, and models. “Two out of the three members of this Llama3-V team are merely undergraduate students at Stanford University, and they have a long journey ahead. If they can acknowledge their mistakes and make amends, it would be a great virtue,” he said.
• Christopher David Manning, the director of the Stanford AI Laboratory, also issued a statement condemning the act of plagiarism and praising the Chinese open-source model MiniCPM.

Context: Founded in August 2022, ModelBest Inc secured a new round of financing worth hundreds of millions of RMB in April. Huawei’s Hubble Technology Venture Capital led the investment, with participation from Chunhua Capital, Beijing Artificial Intelligence Industry Investment Fund, and Chinese Quora-like platform Zhihu. In February, ModelBest Inc launched the open-source model MiniCPM.

Tencent on Thursday launched an AI assistant app called Yuanbao, signaling the Chinese tech giant’s belated entry into the highly competitive artificial intelligence assistant field with an independent entry based on the company’s Hunyuan foundation model.

Yuanbao integrates artificial intelligence-based search, summaries, and writing competencies, and can be used to analyze documents, and basically engage with people based on prompts.

Why it matters: The opening of the independent app shows that Tencent has finally chosen to face competition head-on, several months after AI offerings from its rivals. Baidu’s ERNIE Bot has already amassed 200 million users, while ByteDance’s Doubao currently boasts over 26 million monthly active users.

Details: Tencent’s Yuanbao distinguishes itself from Baidu, Alibaba, and ByteDance’s products by offering direct access to content from the WeChat ecosystem, the company’s super app with nearly 1.4 billion monthly active users, which the firm says will give it superiority in terms of content timeliness and richness.

• WeChat, the all-in-one app, features a fan page functionality similar to Facebook’s, known as Official Accounts, where individual creators or enterprises can publish articles or information to subscribed users. The content cannot be directly accessed through mainstream search engines such as Baidu.
• Since its launch in September 2023, the parameter scale of Tencent’s Hunyuan large model has been upgraded from hundreds of billions to trillions, according to Tencent, and the pre-training corpus has increased to 70 trillion tokens.
• Based on the published guidelines of the Yuanbao app, the consumer-oriented assistant currently offers multiple AI agents, similar to GPTs, which can be selected for various functions, including English-speaking practice, daily life translation, and avatar generation.

Context: Last week, Tencent announced that it had made the lite version of its Hunyuan model free to access, while the price of its more powerful version has been reduced by 50% to 87.5%, as the firm joined a fresh price war among major Chinese tech companies.

• Tencent’s announcement comes just hours after iFlytek said its SparkLite API would be permanently free. In an interview with TechNode at the BEYOND Expo 2024, the firm’s vice president Wang Wei mentioned that the decision was made because “iFlytek should have a sense of social responsibility as a major company by making [AI] technology available for broader applications.”

TikTok owner ByteDance has surprised the artificial intelligence industry with the ultra-low cost of its Doubao model, which the company said is capable of processing 2 million Chinese characters, equivalent to 1.25 million tokens, for RMB 1 ($0.14). OpenAI’s most advanced multimodal model, GPT-4o, also unveiled this week, comes in at $5 per million input tokens handled.

“A good model works well and is affordable to everyone,” Tan Dai, president of Volcano Engine,  ByteDance’s cloud computing services unit, said at an event on Wednesday. “Large usage polishes a good model and dramatically reduces the unit cost of model inference,” he added.

Why it matters: The radical pricing, which Tan explained is achieved through model structuring and hybrid scheduling of cloud computing power, is likely to initiate a price war in China’s AI field after a battle on parameters.

Details: ByteDance’s Doubao large model family consists of nine versions, ranging from a “lite version” to the advanced Doubao Pro, which is able to deal with up to 128,000 token inputs. The range also includes models centered on generating pictures and virtual characters.

• The company did not disclose how many parameters it used to train its most powerful Doubao Pro model.
• The Doubao LLM family shares the same name as ByteDance’s ChatGPT alternative, a free AI bot that the company says has 26 million monthly active users.
• The Doubao model is now open to external developers and other interested parties for cooperation via Volcano Engine, though this comes months after tech rivals Alibaba and Tencent gave enterprise customers access to their models last April and September, respectively.

Context: Earlier this year, ByteDance founder Zhang Yiming criticized employees for not being “sensitive enough” to the emergence of new technologies like ChatGPT in an internal meeting, saying that discussions among staff over ChatGPT only started in 2023. His remarks came after the company had already launched the Doubao chatbot in the second half of last year.

The development of modern computer science is marked by humanity’s relentless push towards the boundaries of computing efficiency: new technologies led to new applications, digital services expanded, super platforms emerged, internet traffic surged, energy consumption increased, fundamental technologies advanced, and eventually computational efficiency improved. Over the past 20 years, the global tech industry has largely been going through such a cycle.

Some argue that the path to artificial general intelligence (AGI) may break this cycle. According to the International Energy Agency’s Electricity 2024 report, after globally consuming an estimated 460 terawatt-hours (TWh) in 2022, data centers’ total electricity consumption could reach more than 1,000 TWh in 2026. This demand is roughly equivalent to Japan’s annual electricity consumption. Thus, should the future of AGI come at the expense of our planet?

I believe we cannot assess tomorrow’s challenges with today’s capabilities, but yesterday’s practices can inspire us and instill some confidence in what we do today. I started working in the tech sector around 2000, and since then, the industry has evolved through eras dominated by mainframe servers, minicomputers, distributed architecture, and cloud computing.

Each iteration of the computational infrastructure is a process of broadening the accessibility of digital services while also representing the evolution of software and hardware technologies to be integrated to improve computing efficiency. What remains constant, regardless of how business demands or computing methods evolve, is the industry’s unchanging pursuit of lower energy consumption and higher efficiency for computing tasks. Since the early days of Alipay, in our pursuit to strike a delicate balance among business expansion, continuity, and cost management, “green computing” emerged as a viable technology solution. 

Green computing is essentially a process driven by technology to sustain business viability. On the path to AGI, the exponential growth in energy consumption is now “the elephant in the room.” However, I believe the tech industry is fully aware of the severity and will, as always, drive innovations in software and hardware, as past breakthroughs have often emerged from seemingly impossible challenges.

Taking Ant Group’s experience as an example, during the “11.11 Global Shopping Festival” of 2010, our payment processing service was only four seconds away from crashing under peak loads. This compelled us to transition from relying on minicomputers to deploying distributed architectures to enhance computational efficiency. By 2021, we implemented green computing technologies at scale, using technologies like workload colocation, cloud-native time-shared scheduling, and AI-based auto scaling, which doubled our CPU server utilization compared to 2019. As a member of the technology team that made this possible, the most visible change for me was that we started from being on high alert in the project room, to eventually enjoying a can of Coke and letting the system handle the traffic peak by itself. 

In the era of intelligent computing, the energy consumption challenge is even greater, something we have experienced firsthand. Just like the 11.11 Global Shopping Festival, large-scale campaigns provide the best testing ground for new technologies. For example, each year, Alipay will launch an annual Chinese New Year campaign called Five Fortune. In 2024, for the first time, we started to pioneer green computing technologies in this AI-powered campaign at scale. 

The 2024 Alipay Five Fortune campaign introduced several AI-powered features, attracting 600 million interactions over 12 days. To scale foundation model applications while controlling computing costs, continuous optimization of hardware performance is required, along with better software hardware integration and algorithmic efficiency improvements. Currently, Ant Group has built a heterogeneous cluster of over 10,000 acceleration cards, where hardware compute efficiency (HFU) exceeds 60%, and the cluster’s effective training duration accounts for over 90% of total time. The RLHF training throughput performance is 3.59 times higher than industry-standard solutions under equivalent model effects.

During these green computing pilots, we have gained two key insights:

• In the era of intelligent computing, companies must incorporate the green computing system into strategic planning from day one. The approach to technology infrastructure is no longer about patching or making incremental improvements to the existing system, but about refactoring. The old method of “letting the business run first, then considering energy consumption” is no longer viable. Companies that realize this sooner will have a head start.

• Green computing is no longer exclusive to large corporations. Previously, it was believed that because computation costs are proportional to scale, large companies had a greater need to drive green computing initiatives, while smaller companies felt less urgency to follow suit. However, in the era of intelligent computing, the high costs of computing power make energy efficiency a critical metric for companies of all sizes and types. Given the barriers to developing green computing and the limited R&D budgets of smaller companies, we believe a green computing market will form in the future, providing products and services to businesses of all sizes.

Looking ahead, as industries undergo digital transformation, a triangular challenge emerges: ensuring the secure flow of data across entities, protecting user privacy, as well as ensuring the business is commercially sustainable. In response to these challenges, we believe the future of smart computing will evolve into the era of cryptographic computing, where data will be processed in an encrypted manner across clouds, regions, and industries. This requires more complex computations and higher computing power, thus making low-cost cryptographic computing essential for companies aiming to capitalize on digital transformation. Ultimately, low-cost cryptographic computing will unlock the value of data circulation for instantaneous use, just like turning on the tap.

From the past era of general computing, through the current era of intelligent computing, and into the future of cryptographic computing, pushing for a green and efficient computational power system has been a common pursuit for the tech industry. We believe that breakthroughs in core technologies that integrate software and hardware are crucial, ultimately achieving an optimal balance of computing power, storage, and networks. To achieve this across sectors, companies need to manage their organizations through data-driven approaches to make targeted optimizations, and actively participate in open-source projects and promote the proliferation and application of green computing technologies.

Problems that arise during technological advancement can only be solved through further technological breakthroughs. Similarly, the issue of energy consumption, exacerbated by technology, can only be fundamentally resolved by technology. When we consider what kind of earth we want to leave for future generations, and when our children ask whether technology can truly bring a better future, the endeavors of our generation in green computing take on a more profound significance.

China’s search engine operator Baidu said Tuesday the number of users of its AI-driven ERNIE Bot has doubled to over 200 million in four months, declaring itself a leader in artificial intelligence, as competitors rush into the sector. The company also added that ERNIE now handles 200 million daily queries.

Why it matters: Baidu’s chatbot has revived confidence in the company’s monetization of AI, with its 200 million users offering new potential avenues for growth. 

Details: Co-founder and CEO of Baidu Robin Li expressed confidence that natural language will emerge as the new programming language in the future in his speech titled “Everyone is a Developer” at the company’s AI Create conference on Tuesday.

• Baidu released three AI development tools during the event – AgentBuilder, AppBuilder, and ModelBuilder, underpinning Li’s view that developing apps should become “as straightforward as making a short video.” According to video demos showcased during the talk, these tools can even help users without any coding skills build apps.
• More than 190,000 AI apps were created through Baidu’s Qianfan cloud platform by over 85,000 businesses, the company claimed in the latest figure.
• In the speech, Li explained the “fine-tuning and post-pre-train methods” based on ERNIE 4.0, the firm’s current most advanced foundation model, enable the creation of models with better results at a lower cost compared to models tuned directly from open source models, which is why Li said he thought open-source models would become “more and more outdated.” His earlier comment that open source models “make little sense,”  recently sparked contention on social media among Baidu’s counterparts. 
• Rival Alibaba developed an open source large language model last August, becoming the first Chinese tech major to do so. Others, including startup 01.AI, founded by Kai-Fu Lee and Baichuan built by Sogou’s founder, both chose to make their billions-parameter models open source soon after.

Context: China’s regulator approved 117 large language models for generative AI services on March 25, according to the Cyberspace Administration of China (CAC). The scope of approvals signals stiff competition among tech companies scrambling to create potential from the technology’s commercial application.

• A recent chart finds Baidu, which was amongst the first batch to open its ChatGPT-like bot to the market last August, is being chased by rival company Moonshot AI-trained Kimi, which specializes in analyzing long text, in terms of total traffic. ERNIE Bot recorded 14.9 million clicks in March, while Kimi’s combined traffic surged by 321.58% to 12.6 million.

Moonshot AI, an artificial intelligence startup founded in China a year ago, has quickly narrowed the gap with rival Baidu’s ERNIE Bot in terms of users after upgrading its large language model Kimi to be able to handle text inputs of up to two million Chinese words in March.

According to a survey conducted by data website AIcpb, Kimi saw its combined website and app traffic top 12.6 million views last month, bringing its ranking up to second place behind Baidu’s chatbot, which received 2.29 million more views. 

Why it matters: The quadrupling of Kimi’s visit count in March meant the ChatGPT-like tool surpassed Alibaba’s competitor Tongyi Qianwen. Its ability to handle long texts is fueling Kimi’s popularity among users, raising hopes among its backers that it will soon break Baidu’s months-long hold on the top position in China’s AI sector.

Details: Moonshot focuses on longer textual inputs and outputs, a new battlefield for Chinese LLM companies. Alibaba opened up a free long-document processing feature that can handle 10 million characters following Kimi’s upgrade, while Baidu plans to support longer text conversations later this month.

• Less than 72 hours after Kimi claimed a technological breakthrough in handling 2 million Chinese characters, the chatbot experienced a service outage. The crash, which lasted at least two days, was blamed on “an abnormal increase in system traffic far exceeding the system’s capacity,” by the company, as users rushed to try out the new feature.
• In a recent interview with Tencent Technology, Moonshot founder Yang Zhilin noted that a focus on long texts was decided upon at the company’s inception. “It’s fundamental,” Yang said, noting that the core value of AI lies in personalized interaction, best achieved, he believes, not through fine-tuning but by supporting broader contexts.
• He also said that while there is “significant room” for expanding text length, focusing on the numbers alone is “meaningless.” Instead, reasoning, faithfulness, and clear instruction-following are what matters.

Context: Moonshot AI reportedly raised over $1 billion in funding earlier this year from existing shareholders such as Alibaba, Sequoia China, and Xiaohongshu, marking the largest single-round financing for a Chinese AI big model company to date. The company raised nearly RMB 2 billion ($277 million) last year.

Baidu has reportedly won a deal to power iPhones and other Apple devices in China using its generative artificial intelligence, domestic media outlet China Star Market reported on Monday, citing unnamed sources familiar with the matter. The Apple tie-up is a potentially major boost to the Chinese search engine giant’s vision of extra AI-generated revenue. 

Why it matters: China’s regulations have pushed Western tech companies to fully comply with the country’s rules if they want to lure the Chinese market, and that benefits Baidu, which serves many of the functions that Google does in the US and Europe. Following the Beijing headquartered firm’s rapid rise to prominence in the early 2000s, it appears poised to gain a major advantage from restrictions on Google once more.

Details: Apple previously engaged in discussions with Alibaba and another unnamed Chinese AI company, according to the China Star Market report, before finally tapping Baidu as its partner for AI services for its upcoming iPhone 16.

• Spurred by the news, Baidu’s Hong Kong-traded stock quickly rose over 6% ahead of the lunch break on Monday, although these gains were erased by more than half by the market close.
• On Weibo, China’s X-like microblogging platform, opinion was divided regarding Baidu’s AI tech being integrated into Apple devices. While some said the news meant that their interest in purchasing a new iPhone was lost instantly, others claimed it showed how good Baidu’s technology is.
• Baidu didn’t immediately respond to a request for comment from TechNode.
• The Chinese launch of Samsung’s Galaxy S24 series integrated Baidu’s AI chatbot Ernie to power its AI features such as content summaries and translations, but the collaboration has drawn scorn on social media. The device’s “Circle to Search” function, which is supposed to give detailed information about a certain part of an image or video that users long-press and circle, has been the subject of repeated complaints from users.
• In one viral example, a user on the Instagram-like social platform Xiaohongshu shared a video where they take a landscape picture and circles a reef in the center of the photo.  Baidu’s Ernie-driven AI provides a shopping link from an e-commerce platform, while Samsung’s Hong Kong version – powered by Google – displays the exact location of the image.

Context: Beijing-based Baidu, seen as China’s AI leader, is among the companies hoping to turn the much-hyped technology into a fresh revenue engine. Its chief executive Robin Li has said that  generative AI could add several billion yuan in incremental revenue for the firm in 2024. In the fourth quarter, Baidu’s AI-related revenue stood at RMB 656 million ($91.0 million).

Note: The article was first published on TechNode China written by Penghui Li and translated by Zinan Zhang and Jake Newby.

Attempting to summarize the current state of the international tech industry is a tricky business. On the one hand, venture capital is down and layoffs are considerably up; on the other, there is huge excitement within the industry at the potential of new technologies such as generative AI.

According to data from Crunchbase, during 2023, total global venture capital is expected to reach its lowest level in five years at $285 billion, a 38% decrease from the $462 billion seen in 2022. Meanwhile, layoffs are also occurring more widely. Statistics show that the number of redundancies at global tech companies increased by 59% in 2023 compared to the previous year, affecting 262,700 people.

Yet with the rise of artificial intelligence technology represented by ChatGPT, the industry is also awash with unbridled excitement in certain quarters. Data indicates that the total financing amount for global artificial intelligence startups reached nearly $50 billion in 2023, a 9% increase over the previous year. Among them, just three companies – OpenAI, Anthropic, and Inflection AI – raised $18 billion.

It’s not only artificial intelligence. In industries such as augmented reality, brain-machine integration, and quantum computing, the technology field is also showing significant progress. From this perspective, as a home to global cutting-edge industries and an important springboard for corporate globalization, London has become a key focal point for tech companies.

London: A new hotspot for technology and climate innovation

Despite a particularly challenging period for the tech sector, London is still full of bright spots, according to Zhao Bingbing, Chief Representative of Greater China at London & Partners.

Zhao points to the GenAI boom as an example of the UK capital’s status. In recent years, investment in generative artificial intelligence companies in London has exploded. In 2023, the total financing of GenAI startups headquartered in London reached $232 million, 20 times the total financing amount in 2020. Among them, AI video company Synthesia and AI decision-making company Quantexa have joined the unicorn camp.

At the same time, established companies in the artificial intelligence field are also turning their attention to London. In June of last year, OpenAI, the developer of ChatGPT, chose London as the location for its first overseas office. OpenAI emphasized that the British city has a “vibrant technology ecosystem” and “outstanding talent.” CEO of Open AI, Sam Altman, further highlighted, “We see this expansion as an opportunity to attract world-class talent and drive innovation in AGI development and policy.”

Beyond AI, climate technology has become another highlight of London’s technology industry. Zhao states that London has always been an advocate and leader in the sustainability field. The UK capital set a goal a few years ago to become a zero-carbon city by 2030 and a zero-waste city by 2050. With strong policy support, “London has become the ideal ‘experimental field’ for international scientific and technological innovation companies in the sustainable technology field,” Zhao says.

London’s climate technology companies raised a total of $3.5 billion in 2023, a significant increase from the $2.2 billion amassed in 2022. Zhao emphasized that last year, London & Partners helped 100 scientific and technological innovation companies from around the world land in London, of which 23% were in the climate technology field, doubling the number who they helped bed in 2022. These companies created nearly 1,000 job opportunities, a 200% increase compared to the previous year. The firms included BYD, the world’s largest electric vehicle manufacturer, which also chose London as its UK headquarters.

From garnering attention to leading change

London’s attractiveness as a tech hub has seen funds pour into the city from around the world.

London startups raised over $12.9 billion in venture capital in 2023, the highest of any European city and almost equivalent to the total of the following three European cities (Paris, Stockholm, and Berlin). This was part of a trend that has seen London’s VC investors raise more than $24 billion in the past three years, accounting for over 30% of the total VC funds in Europe over that period. In addition, London continues to welcome the entry of many influential investment funds including A16z, Antler, BDA Partners, as well as Sustainable Ventures, Octopus Ventures, and 2150, while global venture capital companies such as Index Ventures, Lightspeed Venture Partners, and Sequoia have recently set up important institutions or new funds in the city.

Zhao says that London Tech Week, held annually in June, has played an important role in building the UK capital’s technology ecosystem. The event brings together leading companies, founders, investors, government agencies, and future talents from around the world to discuss the most urgent issues facing business and society.

Zhao’s firm, London & Partners, has long provided diverse community activities for companies registered in London, helping companies – especially small and medium-sized scientific and technological innovation companies – better integrate into the city’s mature business ecosystem and connect with broader resources. Zhao emphasizes that London is increasingly becoming a gathering center for major companies and institutions.

London & Partners have organized community activities like “Meet the Investor” and “Meet the Corporate” to connect scientific and technological innovation companies seeking financing with investment opportunities, or to provide platforms for companies to connect with institutions and large enterprises. Recently, the firm has organized an opportunity for financial technology companies to exchange ideas with the Financial Conduct Authority (FCA) in London. Zhao notes that “In London, our community activities also emphasize internationalization. In addition to our Grow London Global program to support the internationalization of enterprises, the London Tech Week in 2023 attracted participants from 168 countries and regions, as well as many national pavilions — such as from the Netherlands, Germany, Dubai, Poland, and Pakistan — showcasing the development of their respective scientific and technological ecosystems.”

Since its founding in 2014, London Tech Week has helped the UK capital’s technology ecosystem develop into an important center for nurturing startups and growing companies. According to data from Dealroom, the value of London’s technology ecosystem has grown enormously over the past decade, from $70 billion in 2014 to $621.5 billion in 2023. At the same time, since the founding of London Tech Week, London’s technology companies have raised a total of $107 billion.

Helping Chinese companies continue to tell good stories on the international stage

From high-tech products such as robots, EVs, and drones to new overseas territories represented by Latin America, the Middle East, and Africa, from the sweeping presence of Shein in the global fashion retail landscape to the rise of MIXUE Ice Cream & Tea in Southeast Asia, there’s a new wave of companies looking to go overseas.

In this context, it is not hard to see the importance of London as a global springboard for companies that have already joined this journey or are considering including overseas markets in their future expansion plans.

Zhao states that in the past year, London & Partners has helped 20 science and technology enterprises of different scales from the Greater China region to settle in London. In addition to BYD, the world’s largest electric vehicle manufacturer, the agency has provided assistance and support for PingPong Payments, a financial technology enterprise headquartered in Hangzhou, to expand its UK business. From an industrial perspective, these Chinese enterprises come from different industries. London & Partners have also worked with firms from the creative technology, industrial digitalization solutions, and biopharmaceutical industries.

“2023 was a fruitful year for the Greater China region,” says Zhao. “In 2024, we hope to continue to assist more Chinese enterprises planning to go global, enabling them to better integrate into the science and technology ecosystem of the UK and London, benefit from London’s mature and well-established international business ecosystem, smoothly settle in London for development, and expand internationally from London as a starting point. At the same time, we will continue to seek suitable platforms for Chinese enterprises, such as London Tech Week, the GROW Summit, and other technology or industry-related events, to help Chinese enterprises tell their stories on international platforms.”

Baidu co-founder Robin Li has used the example of a Tang dynasty poetry meter to demonstrate what he says is ERNIE 4.0’s superiority to OpenAI’s GPT-4 on Chinese language tasks. In an interview broadcast by Chinese state television last week, the Baidu chief executive said the company’s latest AI chatbot had shown higher aptitude when tasked with creating poetry in the form of a complex Tang dynasty scheme known as Qinyuanchun.

“If I asked GPT to compose a poem following the Qinyuanchun scheme, the tool would become totally confused,” Li said, “because it lacks the understanding of whether the first sentence should consist of four words or five.”

Why it matters: Li’s attempt to emphasize ERNIE’s competitive advantage in certain areas follows a raft of Chinese AI ventures receiving outside capital injections from Baidu’s rivals. His comments also come as OpenAI’s text-to-video model Sora kicks off a new round of AI mania worldwide.

Details: In the same interview, the CEO did acknowledge that ERNIE lags “a little bit behind” GPT in understanding English, as it hasn’t been trained on as much English language data.

• However, comparison is inevitable, and Chinese companies focusing on AI products have always benchmarked their products against OpenAI’s models. Such comparisons have often drawn scorn on Chinese social media, with Baidu itself often suffering in comparison to ChatGPT’s innovative progress. But Li told the CCTV show that he is “not angry” about the stark differences between ChatGPT and his firm’s self-built chatbot, adding that he hopes to change people’s perceptions within a year.
• Li also stated that, in his opinion, the profession of “programmer” will essentially cease to exist in the future as “everyone will possess programming abilities as long as one can speak.” English and Chinese will be the only two programming languages, Li predicted.
• Li also advised companies and entrepreneurs in the market to avoid engaging in the constant launch of large language models, which he regards as “repetitive labor.” He emphasized that “only applications truly create direct value.”

Context: Baidu’s investment in AI has yielded results in its financial reports, with its AI models and ERNIE chatbot contributing RMB 656 million ($91.5 million) to the firm’s revenue in the fourth quarter last year. However,  such contributions amounted to just 1.9% of the company’s total earnings.

• Li was speaking ahead of the one year anniversary of Baidu’s ERNIEbot, which launched on March 16 2023.

Alibaba is reportedly joining as a lead investor in the latest funding round for Chinese AI startup MiniMax, the second major investment deal that the e-commerce giant has made in the emerging sector in 2024.

China Star Market, the local media outlet that first reported the news, noted that the financing has yet to be finalized, while a Tuesday report by Bloomberg said that Shanghai-based MiniMax is seeking at least $600 million.

Why it matters: Despite Alibaba’s cloud computing department’s focus on its own large language model, Tongyi Qianwen, the company remains actively engaged with external investments, casting a wide net as it looks to increase its bets on potential leaders in the country’s artificial intelligence sector.

Details: Minimax, founded by former vice president of SenseTime Yan Junjie in December 2021, has already launched services akin to ChatGPT. It also provides various LLM capabilities for Chinese word processing software WPS and the lifestyle-sharing platform Xiaohongshu.

• The move comes after Alibaba led fundraising for Moonshot AI earlier this year, raising over $1 billion and making it the latest AI unicorn in China. In the record funding deal for Moonshot AI, Alibaba not only supported most of the funding but also contributed through computing power services. This move echoes Microsoft’s investment in OpenAI, where a significant portion was allocated through cloud computing services, according to CNBC.
• Two of China’s leading artificial intelligence startups – Zhipu AI and Sogou founder-built Baichuan AI – also received funding from Alibaba in 2023. 

Context: “Artificial intelligence” was mentioned three times in this year’s government work report, delivered by China’s Premier Li Qiang on Tuesday. Li highlighted ongoing innovations in the field since 2023, while also stressing the need to “deepen research and development applications of artificial intelligence” this year, signaling the Chinese leadership’s desire to accelerate the development of “new productivity”.

• Representatives from the technology sector also discussed AI in their proposals for the so-called Two Sessions, the Chinese government’s annual plenary meetings. For instance, Lei Jun, Chairman and CEO of Xiaomi, suggested integrating AI literacy education into the nine-year compulsory education curriculum and introducing AI general education courses.

A 48% drop in net profit from a year earlier overshadowed Baidu’s record revenue in the fourth quarter of 2023, which was led by advertising income, but also featured a minor contribution from earnings related to artificial intelligence. Baidu’s Hong Kong-traded stocks have plunged nearly 30% since last July.

Why it matters: A 6% year-on-year increase in advertising revenue is the main driver of the search engine giant’s revenue growth, as Baidu’s hopes that AI will serve as a significant revenue stream remain unconvincing to investors, in contrast to US tech firms that saw a share uptick based on an AI boom last year. 

Details: Baidu’s AI model and ERNIE chatbot added RMB 656 million ($91.2 million) to revenue in the fourth quarter of 2023, CEO Robin Li told investors on the firm’s earnings call, with incremental revenue expected to jump to several billion yuan in 2024.

• Li revealed that inference costs for the basic free-to-use ERNIE Bot 3.5, used to analyze how much computing is needed to generate output based on prompts,  are down 99% compared to last March. 
• ERNIE Bot 4.0, the paid version rolled out in October, remains Baidu’s most advanced generative AI bot. The chips Baidu has on hand are expected to advance the flagship bot to the “next level,” said Li.
• Baidu’s increased investment in and commitment to AI is also reflected in the fact that its R&D costs climbed 11% to RMB 6.3 billion last quarter. The company released data showing that its homegrown AI service has attracted a sizable number of business customers, reaching 26,000 as of December.

Context: Baidu’s ChatGPT-style service has recorded more than 100 million users given its first-mover advantage and integration with China’s largest search engine,. But competition has increased, as Baidu’s tech counterparts in the country have also bet strongly on generative AI and there are an increasing number of AI startups that have received large amounts of funding from its rivals, such as Moonshot AI.

Chair of iFlytek Liu Qingfeng told employees that the Chinese voice-recognition company has “resisted the extreme pressure from the US” and is entering “a new period of strategic opportunity,” at his firm’s annual general meeting. Speaking on Tuesday, Liu again benchmarked iFlytek’s upgraded Spark against GPT-4 Turbo, saying it expects to be as powerful as OpenAI’s most advanced model in the first half of 2024.

Why it matters: Liu’s confident tone throughout his speech is being seen as a sign that the company seems to have passed its dark days in the grip of US sanctions due to its “seizing of the strategic opportunity of large language models.”

Details: In his speech, Liu stated that the top goals for iFlytek in 2024 would be to be the fastest Chinese company to catch up with GPT-4, to “empower 10 million intelligent hardware terminals”, to produce “industry-leading applications taking the largest market share in six industries including education and finance”, and to “amass one million LLM developers to build an AI model ecosystem”.

• Liu’s address on Tuesday emphasized the importance of building “independent and controllable” home-grown AI foundation models, with the exec saying LLMs are a battlefield for gaining strategic advantage over the US.
• However, the company’s financial forecast offers a glimpse into the huge cost of achieving China’s ChatGPT alternative, with iFlytek’s net profit excluding extraordinary profit and loss expected to be down 71% to 81% to between RMB 80 million and RMB 120 million ($11.3 million to $16.9 million) in 2023 compared to the previous year iFlytek has claimed this is largely due to “firm investment” in core technology and industrial chain localization substitution.
• iFlytek also projected an up to 7% increase in revenue to RMB 20 billion for its financial year ending Dec. 31.
• iFlytek unveiled its Spark iteration model on Jan. 30, claiming it as the first such model to be built on the AI model training platform the company designed in cooperation with Huawei and which uses the Chinese telecom giant’s AI chips.

Context: The US government blacklisted A-share listed iFlytek in Oct. 2019, meaning China’s top artificial intelligence company was banned from buying critical components from US companies without US government approval.

Note: The article was written by Steven Lee and translated by Zinan.

Get ready for the annual insights from TechNode Content Team! The year 2023 can be considered a groundbreaking year in the technology field. As wrapping up this year, we gathered different insights from our content team. We’ll present nine Q&As, with timely updates every Wednesday and Friday in the following weeks!

Today, our Q&A comes from Steven Lee, Editor at TechNode. Steven is an editor who focuses on Apple, Google, Microsoft and other big company dynamics.

1. Which company has impressed you the most in 2023?

This year, the most impressive company for me is OpenAI. ChatGPT is the most eye-catching product of the year, sparking a trend in AI. This generative AI has influenced various fields, including creativity, design, media, etc. Tech giants in China and the U.S. are competing to develop and launch similar products.

2. Which company has surprised you the most in 2023?

TikTok Shop being shut down in Indonesia by the government was unexpected. The Indonesian government believed that TikTok Shop’s business model would impact the livelihood of local small businesses, leading to the shutdown order. Later, TikTok announced a $1.5 billion investment in the Indonesian e-commerce platform Tokopedia. The developments of these two companies will continue to be watched closely.

3. Which industry professional/entrepreneur/startup founder has left the most profound impression on you in 2023?

Sam Altman, the creator of ChatGPT. After being removed from the CEO position, Microsoft announced his hiring along with co-founder Greg Brockman. In a surprising turn of events, with the support of core employees and investors, Altman and Brockman returned to the company, highlighting their influence and status.

4. What is the most memorable overseas event for you in 2023?

Sam Altman, the creator of ChatGPT, was unexpectedly removed from the CEO position by the OpenAI board. Later, Microsoft announced his inclusion, and he returned to OpenAI in a strong comeback.

5. If you were to recommend one significant industry trend for everyone to follow, what would it be?

NVIDIA’s collaboration with Malaysia’s YTL Group to build a $4.3 billion AI data center in Malaysia. The development of NVIDIA in Malaysia is anticipated, and there are hopes that it will drive the growth of AI in the country. Due to the popularity of ChatGPT, AI has become a focal point, making NVIDIA the best-performing listed company on the S&P 500 index last year. The founder, Jensen Huang, created a buzz with his whirlwind tour of Southeast Asia, including Singapore, Malaysia, and Vietnam, at the end of last year.

6. What industry buzzword have you encountered the most in 2023?

Artificial Intelligence (AI), ChatGPT.

7. Which phrase or sentence best summarizes your perspective on the field you’ve been following in 2023?

AI field: Rapid advancement, intense competition. With the significant success of ChatGPT, major tech giants and startups are competing to launch AI-related products, leading to a rapid advancement of AI.

8. What product/company/technology/industry are you most looking forward to next year?

I’m very excited about the developments in the Electric Vehicle (EV) sector. Huawei and Xiaomi have recently launched their EV models, and I’m eager to see the impact and changes they will bring to the overall EV industry. Additionally, the Chinese EV brands BYD and NIO have surpassed Tesla in global sales, and I’m looking forward to the plans and developments of BYD, especially its introduction of products in Indonesia in January.

9. Do you believe AI has the potential to threaten humanity?

I believe AI may affect some jobs, but currently rather than replacing people’s jobs, I tend to believe that AI is helping people enhance efficiency and solve problems such as quick data collection.Furthermore, I have confidence that humans will find ways to control or balance the threats posed by AI and use AI more effectively to enhance efficiency and solve various problems.

Note: The article was written by Angelo.

Get ready for the annual insights from TechNode Content Team! The year 2023 can be considered a groundbreaking year in the technology field. As wrapping up this year, we gathered different insights from our content team. We’ll present nine Q&As, with timely updates every Wednesday and Friday in the following weeks!

Today, our Q&A comes from Angelo, Editor at TNGlobal. Angelo is an editor who focuses on SEA Tech News.

1. Which technology has impressed you the most in 2023?

Undoubtedly, artificial intelligence is the hottest topic in recent years, and OpenAI has been at the forefront of this industry. It’s not necessarily for its advancements in technology, because other companies like Google, Microsoft, Meta, etc., have also been building upon their own capabilities. Many of these are under-the-radar, but OpenAI accelerated the time-to-market for AI-driven products, for better or worse. More importantly, the management “drama” that unfolded toward the end of 2023 highlighted the importance of governance and accountability in this emerging technology, especially given its potential impact on humanity.

2. Which company has surprised you the most in 2023?

Frankly, Microsoft is the single company that has surprised me this year. While many would view Microsoft as a bland company that now lacks the sexiness and edge that other giants like Apple and Google have, it does have significant investments in products and technologies that can potentially be gamechangers, particularly AI. It is also making significant inroads into community-driven technologies such as open-source projects, which is one of the things I am deeply interested in.

3. Which industry professional/entrepreneur/startup founder has left the most profound impression on you in 2023?

OpenAI’s Sam Altman made an impression on me not only because of AI, but rather because of his interest in survivalism and prepping. Altman reportedly has a bunker in New Zealand, same as many other billionaires who can afford the infrastructure, supplies, personnel, and training involved in prepping for disasters in the long-haul. I am mostly interested in prepping for more localized ones, such as those brought about by natural hazards (e.g., earthquakes, storms, etc.). But given Altman’s involvement in AI, does he know something we don’t? Will AI eventually lead to humanity’s ruin?

4. What is the most memorable overseas event for you in 2023?

I am particularly interested in exploring the unknown and pushing human capabilities and technologies to the extremes. For this reason, I have two events from 2023 that are quite memorable to me. One is India’s successful Chandrayaan-3 mission to the moon, which is perhaps one of the factors that will rekindle renewed interest in moon missions and exploration, not to mention the other activities by NASA, SpaceX, CNSA, ESA, and other organizations in advancing space-exploration technologies.

Another is the OceanGate disaster, which involves another area of exploration–the deep oceans. With its loss of its Titan and crew, it also underscores the importance of establishing a balance between achieving innovations and developing standards for safety.

We’ve achieved so much in these areas in the past century alone. I’m excited for what we can achieve in the 21st century onwards.

5. If you were to recommend one significant industry trend for everyone to follow, what would it be?

There’s always talk about how electric vehicles will enhance the sustainability of our transportation systems. For me, however, what’s more interesting is the deep technology behind these industries, particularly alternative power sources. Electric vehicles may simply be shifting the carbon footprint from fossil fuels to other sources of energy. But what about new breakthroughs that can potentially be cleaner and safer, such as the use of hydrogen and other materials? And this might not only be for terrestrial travel. How about propulsion or space-going vessels?

6. What industry buzzword have you encountered the most in 2023?

AI, of course, and its related buzzwords, had been the most popular in 2023. Other contenders would be “The Great Resignation” in the context of work-from-home and return-to-office trends, as well as “Open Source Intelligence” in the context of data management.

7. Which phrase or sentence best summarizes your perspective on the field you’ve been following in 2023?

“Anxious but excited,” is the phrase that best summarizes my view especially in tech-driven industries across the globe. On one hand, we are having significant advancements in artificial intelligence, which means we can come at innovations and efficiencies at a scale previously not achievable. However, this also means potential disruption to a lot of industries. Some might get left behind if we are not able to get ahead of the wave.

8. What product/company/technology/industry are you most looking forward to next year?

We have been exploring space for decades. With recent advancements in reusable materials and propulsion technologies, we might just achieve significant developments in the next years to come that will make space travel more accessible and viable than ever.

9. Do you believe AI has the potential to threaten humanity?

I think humanity poses a threat to humanity. AI will just be something that accelerates it if left unchecked. We therefore need better governance mechanisms that will ensure AI is not utilized for harmful applications and that AI itself will not have the means to do such on its own.

Note: The article was written by Evan Huang and translated by Zinan Zhang.

Get ready for the annual insights from TechNode Content Team! The year 2023 can be considered a groundbreaking year in the technology field. As wrapping up this year, we gathered different insights from our content team. We’ll present nine Q&As, with timely updates every Wednesday and Friday in the following weeks!

Today, our Q&A comes from Evan Huang, reporter at TechNode. Evan Huang is a tech reporter based in Shanghai who focuses on consumer electronics, hardware, and AI.

1. Which company has impressed you the most in 2023?

Huawei. Under U.S. sanctions for over three years, Huawei’s business faced setbacks. However, the launch of Mate60 Pro on August 31 sparked a wave of enthusiasm. Huawei, now using self-developed chips again, is considered to break the U.S. technological blockade, showcasing the strength and resilience of Chinese technology.

2. Which company has surprised you the most in 2023?

Apple. iPhone, as a trendsetter in the smartphone industry, experienced minimal overall upgrades in 2023 which needed to be improved. Moreover, it is the first time that Apple hasn’t released any new iPad in the entire year. Since the initial launch of the iPad in 2010, Apple has consistently introduced new iPads each year, making 2023 an exception.

3. Which industry professional/entrepreneur/startup founder has left the most profound impression on you in 2023?

Lei Jun. On November 29th, Lei Jun, the founder, chairman, and CEO of Xiaomi Group, made a personal donation of RMB 1.3 billion to Wuhan University where he graduated from. This is the largest individual cash donation received by any university in China. Lei Jun believes that basic disciplines have been relatively disadvantaged in universities in recent years. Therefore, he donated this amount to support fundamental research in six disciplines (mathematics, physics, chemistry, literature, history, and philosophy), technological innovation in the field of computer science, and the education of university students.

4. What is the most memorable overseas event for you in 2023?

In November, OpenAI CEO, Sam Altman, faced a sudden dismissal, after a series of twists, he briefly joined Microsoft before ultimately returning to OpenAI to reclaim his position. Throughout this process, the major disagreement between Altman and former Chief Scientist Ilya at OpenAI revolved around the pace of commercialization. Ilya advocated for a slower approach, emphasizing the importance of interpretability and safety in AI products. In contrast, Altman and former President Greg held a stance of technological optimism, pushing for an active pursuit of commercialization paths and a new round of financing.

5. If you were to recommend one significant industry trend for everyone to follow, what would it be?

In September, Huawei announced the launch of HarmonyOS applications. It is said that the next-generation, HarmonyOS NEXT system base, is completely self-developed, eliminating traditional AOSP code and only supporting applications based on the HarmonyOS kernel and HarmonyOS system. This new design reduces 40% of redundant code, completely breaking free from the “skin” of Android. At the same time, by focusing on the development of HarmonyOS applications, Huawei can expand the influence of the HarmonyOS system and enhance the independent and controllable level of related technology and ecology.

6. What industry buzzword have you encountered the most in 2023?

Large Language Models. With the rapid development of technologies such as deep learning and big data, significant breakthroughs have been achieved in the training and application of large models. Large language models exhibit stronger generalization capabilities and higher performance, demonstrating outstanding abilities in areas such as speech recognition, natural language processing, computer vision, and more. The industry widely recognizes their enormous application potential in many commercial scenarios.

7. Which phrase or sentence best summarizes your perspective on the field you’ve been following in 2023?

“Seeking Change”. The Consumer Electronics Market is undergoing unprecedented transformation, with various emerging technologies and products continuously emerging. AI has brought a new atmosphere to hardware products in areas such as smartphones and smart homes, providing consumers with more choices and better experiences. At the same time, market competition is becoming increasingly fierce, and manufacturers need to keep pace with the times to meet consumer demands.

8. What product/company/technology/industry are you most looking forward to next year?

AI Agent. The AI Agent will connect to other services and solve practical problems, making the interaction between humans and AI more profound, rather than just simple input-output interactions.  For example, a mature AI Agent can significantly reduce the cost of software production.  Microsoft founder, Bill Gates, has expressed, “Whoever wins the personal agent, that’s the big thing, because you will never go to a search site again, you will never go to a productivity site, you’ll never go to Amazon again.”

9. Do you believe AI has the potential to threaten humanity?

AI is a double-edged sword, and my answer is temporarily uncertain. AI creates more opportunities and possibilities, reshaping global business, technology, education, and other fields. Additionally, with the continuous improvement of AI’s capabilities in decision-making and execution, there may be some unpredictable and uncontrollable situations.   Moreover, the widespread adoption and application of AI may lead to excessive reliance on technology, potentially weakening certain fundamental skills and judgments in humans.   In this process, it is essential to closely monitor and establish appropriate regulatory and ethical frameworks to ensure that the development of AI has a positive long-term impact on humanity.

Chinese tech giant Alibaba is doubling down on artificial intelligence to spur the growth of its e-commerce division, after founder Jack Ma wrote an internal memo saying the era of AI e-commerce “has just begun.” The company’s domestic and international retail businesses are testing multiple AI-driven tools and are on a hiring spree for AI-related roles, according to local outlet LatePost.

Why it matters: Alibaba hopes that by integrating generative AI into its core e-commerce business, it will further boost sales in the highly lucrative sector and create new forms of customer experience to up business retention rates.

Details: Taobao and Tmall Group have channeled their AI focus into four key areas:  marketing platform Alimama, customer apps, merchant apps, and industry-specific apps. Before the reorganization led by Alibaba’s chief technology officer Wu Zeming, there were around 20 teams within the group exploring AI-related businesses, the report said.

• This year’s Singles Day shopping festival, which started in late October, was a testing ground for Alibaba-created AI shopping assistants, including Taobao Wenwen, a chatbot designed for consumer-facing interaction that relies on its pre-trained generative ability to provide users with shopping recommendations. 
• The firm also introduced at least ten free tools for merchants to assist in their marketing and merchandising efforts during the month-long bonanza. While Alibaba kept silent on sales achieved during the festival this year, it announced that the Wenwen chatbot engaged with over five million users, and the merchant tools were used over 1.5 billion times.
• Enabling businesses to achieve lower operation costs through generative AI solutions is seen as a commonality among Alibaba’s AI products. Earlier last month, in a pilot launch event of Aidge, a tool facilitating features from translation and marketing to localized content and design, vice president of Alibaba International Zhang Kaifu said that it is dedicated to “simplifying global operations for small and medium-sized enterprises.” 
• Zhang also leads AI initiatives within Alibaba International Digital Commerce Group, overseeing a team of more than 100, according to LatePost, with one-third engaged in training large language models.

Context: AI appears to be a top priority in Alibaba’s strategy since new management took over in September. In his first letter to all staff, CEO Eddie Wu announced “user first, AI-driven” as his vision for the 24-year-old company, a business that has recently been overshadowed by rivals. “We will realign our operations and refocus our business around these two core strategies,” said Wu.

• An employee posted a critical note on an Alibaba internal forum after Temu parent PDD closed in on  the company’s market valuation late last month, a move that prompted Jack Ma’s rare appearance with him calling on staff to provide more constructive comments and suggestions, especially innovative ideas to the group.

Baidu said on Tuesday it has “comprehensively reshaped” its product portfolios thanks to its AI foundation model ERNIE, resulting in increased operational efficiency, after the Chinese search giant company reported 6% year-on-year revenue growth to RMB 34.4 billion in the third quarter. However, the company’s cloud unit reported its first decline in revenue in almost three years.

Why it matters: Baidu pointed to artificial intelligence bringing about positive changes in key performance indicators across all of its businesses in its latest earnings report. The firm also highlighted that its ChatGPT-like service ERNIE Bot has already amassed 70 million users.

Details: Although the company’s overall growth was up, Baidu’s AI cloud revenue decreased by 2% in the three months to September. Robin Li, CEO of the firm, said this was due to weak demand in smart transportation projects, despite strong demand for generative AI. The figures marked the first decline in Baidu’s cloud business since it started announcing its earnings separately in the last quarter of 2020.

• Online advertising business, which accounts for over half of Baidu’s total earnings, experienced 5% year-on-year revenue growth during the period. This growth rate was slower than the preceding two quarters and lags behind some of its internet-focused peers.
• During the related earnings call, Li expressed confidence that Baidu’s current chip reserves could “keep improving ERNIE Bot for up to two years” amid the US’s tightening of curbs on advanced chip exports to China. He did, however, add that Baidu will still seek alternatives due to the pace of AI development in China being “inevitably” impacted by difficulties in acquiring the most advanced chips.
• The chief executive of Baidu warned a week ago that China’s rush to develop multiple large language models had led to a “wasting of resources,” and he again told investors on the call that “the best option on the market” is companies developing AI-native apps based on Baidu’s foundation model rather than training models themselves.
• Baidu also named JD CEO Sandy Xu as an independent director of its board effective from the beginning of 2024.

Context: Baidu’s large language model, ERNIE, was made available to the public at the end of August after receiving government approval. Last month, Baidu launched the latest version of its AI model, ERNIE 4.0, seven months after the release of its first version.

On Tuesday, US chip giant Nvidia revealed that the US government has instructed it to immediately halt the export of certain high-end artificial intelligence chips to China, as regulators have expedited the enforcement of new restrictions, which were originally scheduled to take effect from Nov. 16.

Why it matters: The US withdrew Nvidia’s 30-day exemption period for chip exports to China on Oct. 23, implementing the new regulation 24 days earlier than expected. Currently, Chinese customers, such as Tencent and ByteDance, can no longer obtain any AI-related products from Nvidia. 

Details: The US government notified Nvidia of the immediate implementation of export restrictions on AI chips starting Oct. 23, as per Nvidia’s announcement to the US Securities and Exchange Commission. The affected products include five GPUs (graphics processing units): A100, A800, H100, H800, and L40S.

• On Oct. 17, the US Department of Commerce announced a series of new restrictions on chip exports, addressing loopholes identified after the US imposed export limitations on chips last October. These new regulations broaden the definition of advanced AI chips and impose additional licensing requirements on chip products destined for over 40 countries and regions, aiming to prevent resale to China. According to the regulations, the new restrictions will take effect from Nov. 16.
• Following the implementation of the new regulations, Nvidia must cease shipping A800 and H800 chips to China unless it has express permission from the US government. A800 and H800 chips are alternative solutions Nvidia offered in place of the originally prohibited A100 and H100 chips, following the initial AI chip export restrictions imposed by the US last October. The Nvidia L40S, an advanced GPU for data centers, will also be affected by the new restrictions.
• Due to the widespread global demand for these products, the sudden acceleration of US restrictions is not expected to have a short-term impact on the company’s financial performance, Nvidia said in its announcement. 
• However, Nvidia CEO Jensen Huang earlier stated that the new ban is expected to significantly impact Nvidia’s sales in the Chinese market, though he added that the company remains committed to complying with US regulations. 
• Several AI industry professionals in China have voiced concerns and doubts over the new measures, according to a report by local media outlet TMTPost. They worry about the potential impact of the new restrictions on the future training of large-scale AI models in China, which could result in a tech gap compared to US-based AI companies like OpenAI.

Context: In August, Nvidia reported a fourfold increase in its Data Center revenue over the last two years, establishing itself as a leader in AI chips with a market share of over 70%. Nvidia’s stock continues to rise, as the company achieved a market cap of $1 trillion earlier in 2023.

• The data center sector comprises processors such as central processing units (CPUs), data processing units (DPUs), and graphic processing units (GPUs). GPUs are favored for AI applications due to their ability to handle multiple tasks simultaneously.

Baidu released the latest version of its AI foundation model ERNIE on Tuesday at the Baidu World 2023 conference, a mere four months after the previous release. At the launch, the tech giant claimed the capabilities of ERNIE 4.0 were as advanced as OpenAI’s GPT-4 model.

Why it matters: Baidu has emerged as one of China’s fastest companies in leveraging AI models to transform existing products, ensuring its prominence in the competitive space.

Details: Co-founder and CEO Robin Li showed off ERNIE 4.0 during a one-hour presentation, showcasing the model’s ability to generate advertising posters and marketing videos in real time. Li also asked it to write a martial arts novel based on prompts, underscoring the enhanced memory capabilities of the iterated model.

• Baidu did not disclose the number of parameters used to train the updated model at the event, merely saying that ERNIE 4.0 saw “similar improvements” in understanding and prompting capabilities, but without specifying whether this was compared to the previous iteration or to some other standard. The company said that enhancements in memory and reasoning are twice and three times that of understanding, respectively, but again didn’t clarify to what they were comparing the new model.
• The search giant has embedded AI capabilities into its flagship products including search and maps, introducing new upgrades in recent months. Li laid out his vision for Baidu, aiming to leverage AI-driven thinking to create native apps tailored for the AI era. He singled out Baidu Wenku, an online interactive document-sharing platform, as a product that has undergone radical transformation, evolving into a “productivity tool” that assists users in content creation.
• Baidu failed to launch new official plug-ins for its chatbot service ERNIE Bot this time, but the company mentioned that around 27,000 developers have applied to join the firm’s plug-in ecosystem platform since it was unveiled last month. The company first incorporated plug-in functionality into the ChatGPT-like service when it upgraded ERNIE to version 3.5 in June, and then added three more plug-ins two months later. 

Context: Rebuilding applications with large language models has been embraced by tech giants from Microsoft to Baidu and Alibaba. Robin Li sees the AI model as an opportunity to overhaul all of Baidu’s products, aligning with the vision of Daniel Zhang, former chairman and chief executive of Alibaba, who also stated that all Alibaba products would undergo a comprehensive upgrade through integration into its AI model.

• ERNIE Bot, powered by Baidu’s foundation model, was made fully accessible to the public on August 31, and has amassed a user base of 45 million, according to the company’s announcements on Tuesday.

Baidu on Saturday appointed its chief information officer Li Ying to lead the company’s AI speaker subsidiary, Xiaodu Technology, after the sudden resignation of former chief executive Jing Kun.

Why it matters: As the Vice President of Baidu, Li’s taking over as the top executive of Xiaodu may signify the strengthening of the company’s ERNIE Bot support for various artificial intelligence product lines under Xiaodu, with the tech giant betting big on its subsidiary’s potential for transformation in the AI field.

Details: Officially launched in 2015 as a smart life business group under Baidu, Xiaodu was spun off from the search giant in 2020.

• Xiaodu’s former CEO Jing Kun recently resigned for personal reasons, according to local media outlet Caixin. The report also noted that Jing was originally scheduled to give a speech entitled “The era of large models, Xiaodu reshaping smartliving” at Baidu’s annual event, Baidu World 2023, on Oct. 17.
• His successor, Li Ying, joined Baidu in 2004. The long-term Baidu employee has been involved in various businesses within the company, including natural language processing, complex search, and Baidu Maps, publicly available information shows. Li will report directly to Baidu’s chairman Robin Li.
• Before her new appointment, Li also led the rebuild of Baidu-developed instant messaging software product Infoflow based on the firm’s ChatGPT-like AI model, according to tech outlet 36Kr. The software is capable of performing AI-driven tasks such as meeting summary generation and intelligent coding after integrating Baidu’s ERNIE Bot.

Context: In May, Xiaodu teased its first smartphone, Qinghe, which is intended as a child-focused smartphone and features English-speaking and location tracking functions. Although the independent Baidu subsidiary stated on its official WeChat public account in February that Xiaodu-created devices would integrate all the capabilities of ERNIE Bot, Qinghe ultimately utilized “a large AI model for studying” developed by the company itself.

• In January, Xiaodu completed a Series B+ financing round participated in by the Chinese state-backed enterprise restructuring fund, with the company’s valuation reaching RMB 35.5 billion ($4.9 billion).
• More than 40 million households already use Xiaodu’s products, including speakers, televisions, as well as other smart home appliances, according to the company.

TSMC is urgently seeking equipment suppliers from whom it can buy CoWoS (Chip on Wafer on Substrate) machines, as Nvidia, AMD, and Amazon expand orders for AI chips, local media outlet Economic Daily News exclusively reported on Monday. TSMC has increased its equipment orders for CoWoS by 30% to meet growing AI-fueled demand, the report claimed.

Why it matters: The AI boom has reshaped the semiconductor landscape, positioning chip-making companies as critical enablers of the tech revolution while presenting complex challenges related to supply chain resilience and technological advancement. 

Details: CoWoS is a high-density advanced packaging technology developed by TSMC for high-performance chips. The current shortage of CoWoS packaging capacity has become the main bottleneck in the production chain for AI chip orders.

• Nvidia is TSMC’s largest customer of CoWoS advanced packaging, accounting for 60% of the production capacity, according to the report. Recently, in response to the strong demand for AI computing, Nvidia has increased its orders from TSMC, while AMD, Amazon, and Broadcom have also been placing urgent orders.
• Due to its shortage of CoWoS machines, TSMC has sought the assistance of local equipment suppliers including Scientech Corporation, AllRing-Tech, Grand Process Technology, E&R Engineering Corporation, and Group Up Industrial. TSMC has increased its original equipment orders by up to 30%, with delivery expected in the first half of 2024. Mass production will commence in the second half of 2024.
• Currently, TSMC’s monthly capacity of CoWoS advanced packaging is approximately 12,000 units, sources familiar with the matter told Economic Daily News. To accommodate the proposed production expansion, the monthly capacity of CoWoS will be increased to between 25,000 and 30,000 units.
• Major customers such as Nvidia and AMD have increased their orders of TSMC wafers in the third quarter, boosting TSMC’s 7nm and 5nm advanced process capacity utilization.
• Pressure on CoWoS production capacity is expected to ease after next summer, TSMC’s president C. C. Wei stated during a recent earnings call. The Economic Daily News report revealed that TSMC has been expanding its facilities at plants including Zhuke, Zhongke, Nanke, and Longtan, so as to increase CoWoS production capacity.

Context: On September 24, Chinese media outlet IThome reported that Qualcomm’s next generation Snapdragon 8 Gen 4 may have been manufactured using TSMC’s N3E process technology, as indicated by leaked documents from Qualcomm.

• On September 8, TSMC announced net revenue for August 2023 of approximately NT$188.69 billion ($5.87 billion), representing an increase of 6.2% from the previous month.

Chinese search giant Baidu launched its ChatGPT-like service ERNIE Bot for public use on Thursday, as one of the first batches of companies given permission to allow regular access to generative AI bots, having filed details of its algorithms with the government. The move signals a softening of Beijing’s regulatory stance towards artificial intelligence.

Why it matters: The approval comes two weeks after China’s new AI rules took effect, paving the way for an initial eight companies to cater their generative AI services to over 1 billion Chinese internet users.

• For China’s dozens of homegrown AI large language models, being among the first to launch could potentially bring early player advantages, given the relatively small distinctions between each consumer-facing service.

Details: The first tranche of approvals has been granted to tech companies and research institutes headquartered in Beijing or Shanghai, from Baidu, ByteDance, and SenseTime to the state-backed Chinese Academy of Sciences and Shanghai Artificial Intelligence Laboratory.

• Local media outlet Beijing News reported on Thursday that, in addition to the first eight entities given approval, Shenzhen-based tech giants Huawei and Tencent, as well as Hefei-founded iFlytek, are readying to unveil their artificial intelligence bots to the general public.
• Alibaba, located in Hangzhou, is not listed on the approved entities, but a source from the company’s cloud unit revealed that its chatbot service, known as Tongyi Qianwen, has completed its filing process and is ready for rollout, according to tech outlet China Star Market.
• Baidu’s ERNIE Bot topped the free app download chart of Apple’s App Store 12 hours after its public availability announcement. The company is gearing up to introduce an array of fresh AI-native apps.

Context: China implemented detailed regulations for generative AI services on Aug. 15, making it clear that government approval of algorithms is a threshold that tech companies must cross before offering AI products to the public, as a way to better control content.

• In an earnings call last week, Baidu CEO Robin Li noted that the government “has increasingly recognized” ERNIE and ERNIE Bot, believing that this endorsement stands the company in good stead for a large-scale release.

Chinese e-commerce giant Alibaba reported a solid quarterly performance on Thursday, attributed to improvements across all of its operations. Notably, its logistics unit Cainiao and digital media arm both achieved profitability for the first time. Total revenue reached RMB 234.16 billion ($32.3 billion), marking a 14% year-on-year increase and a 12.5% rise from the previous quarter.

Why it matters: Alibaba has successfully returned to double-digit growth after a year of relatively stable performance. This achievement comes despite a number of China’s economic indicators suggesting a gloomier overall picture. 

• Revenue growth in the June quarter was primarily driven by Taobao and Tmall, Alibaba’s most profitable businesses, with a significant surge in both the volume and value of orders. China’s second-largest online shopping event, the month-long 618 festival, was also included in this period.

Details: The results of the quarter to June mark the conclusion of Daniel Zhang’s leadership, as he prepares to fully focus on leading the company’s cloud business starting in September. The cloud and computer division experienced the slowest revenue growth (of 4%) among Alibaba’s six independent businesses, while the firm’s international commerce division saw the most rapid expansion with a robust growth rate of 41%.

• Zhang, who is also the CEO of Alibaba Cloud, stated that the 4% growth rate was partly influenced by a “revenue decline from a top customer,” during an earnings call. Nevertheless, he said he believes that the opportunity for AI-driven growth has “just begun.” Alibaba is among the first batch of Chinese companies to launch its own large model and ChatGPT-like tool.
• Despite the robust demand for AI cloud services, the supply of GPUs in the Chinese market fell short, according to Zhang. He added that the cloud and computing unit “only partially met” the demand for AI-related services received.
• Alibaba’s overall net profit experienced a significant 51% year-on-year increase between April and June, while the cloud division more than doubled its profits during the same period.
• Cainiao, which has initiated the process of going public, posted a profit for the first time, making RMB 877 million. This comes after the unit recorded a RMB 185 million loss in the same period last year. The logistics arm’s expansion efforts, both internationally and domestically, contributed to quarterly revenues of RMB 23.2 billion, up 34% year-on-year.
• Alibaba’s digital media and entertainment division also posted its first-ever quarterly profit, spurred by a post-pandemic resurgence in offline shows and cinema screenings. Coupled with an increase in subscription revenue from the streaming site Youku, these incomes helped to boost the unit’s balance sheet.

Context: These are the first quarterly results from Alibaba since the Hangzhou-based powerhouse split its business into six divisions in March. Daniel Zhang will step down as CEO and chairman of the group on Sept. 10, to be succeeded by Eddie Wu and Joe Tsai, who will assume the positions of CEO and chairman, respectively.

Foxconn, Apple’s biggest supplier, has secured an exclusive order to supply Apple’s AI servers, which will ship from the company’s Vietnamese factory, Taiwanese media outlet Economic Daily News reported on Monday. Apple’s demand for AI servers and data centers has increased significantly as the company expands its investment in AI development.

Why it matters: The upsurge of generative artificial intelligence products has swept the technology industry in 2023. ChatGPT, the large language model from OpenAI, was estimated to have 100 million monthly active users in June, making it the fastest-growing application in history. In due course, Apple plans to apply its heavy investment in AI research to products including the iPhone and iPad, according to the Financial Times. 

Details: As Apple’s largest supplier, Foxconn not only assembles iPhones but also acts as a server supplier for Apple’s data center, according to the Economic Daily News report. With Apple’s recent stepping up of its interest in AI applications, the need for Foxconn’s support has increased.

• Foxconn has taken the global lead in the AI server market. The manufacturer already accounts for approximately 40% of the global server market, according to Reuters. Foxconn also provides AI servers for major tech companies including OpenAI, Nvidia, and Google.
• Apple CEO Tim Cook confirmed last week that the company has been studying various AI technologies including generative language models for years. Industry observers have speculated that the tech giant’s increase in AI investment may have contributed to its R&D expenditure reaching $22.61 billion last quarter.
• Apple is reportedly building a large-scale voice model named Ajax and is expected to launch a generative chatbot soon. Given Apple’s large user base around the world, this puts potential pressure on Google and Microsoft as the tech titans also look to launch similar AI offerings.
• There are more than 1.46 billion active iPhone users worldwide as of 2023, according to data from analytics firm Demand Sage. iPhone users account for 21.67% of the world’s total smartphone population, according to the same report. Apple’s mature App Store ecosystem also gives it an advantage over other companies in rolling out AI applications.
• Foxconn’s server revenue reached RMB 1.1 trillion ($138.64 billion) last year. According to industry estimates cited by the Economic Daily News, its AI server-related orders surged by billions of dollars in 2022.
• Apple is recruiting for dozens of AI-related jobs in its offices across California, Seattle, and Paris. New recruits are expected to engage in the research of large language models that can generate text, images, or codes in response to simple user prompts.

Context: Securing an AI server order from Apple has solidified Foxconn’s leading position as a supplier of AI servers. On August 4, Nvidia shifted some of its AI server orders from its manufacturing partner Wistron to a subsidiary of Foxconn, due to concerns over Wistron’s insufficient production capacity.

On Tuesday, Apple removed more than one hundred AI-related apps that offered ChatGPT-style services from its China App Store. In a notice sent to affected developers, Apple stated that the removal was “pursuant to orders by the Chinese government” as the related content is deemed “illegal” in China.

Why it matters: The targeted removal of AI-related apps comes two weeks before the implementation of China’s artificial intelligence regulations. The action may send mixed signals to AI developers who aim to offer AI tools for users in China while striving to comply with the country’s regulations.

• Dozens of generative AI apps were removed in a single day from China’s App Store, data from mobile application analysis platform Qimai shows, including apps powered by ChatGPT and domestic-grown iFlyTek’s AI chatbot SparkDesk.

Details: A source close to Chinese regulators said the reason for taking the apps off the store is that they are not standardized enough in terms of data collection and usage, according to local outlet China Star Market. The source added that it is expected to “take a long time” before these apps are allowed back to the store.

• A developer behind OpenCat, a popular AI app that had a rating of 4.8 before its removal early Tuesday, shared a screenshot from Apple entitled App Review on Twitter, which stated that the Chinese government has tightened regulations over deep synthesis technologies and generative AI. OpenCat, powered by ChatGPT, did not secure a license from the Ministry of Industry and Information, which may have contributed to its removal from the App Store.
• SparkDesk, an AI application developed by voice recognition company iFlyTek and built on its proprietary large language model, has been relaunched on the iOS Marketplace 12 hours after it was taken offline. 
• Search giant Baidu’s ERNIE Bot is still available for download on the Chinese App Store, with only those with invitation codes having access to the chatbot service.

Context: As the latest wave of artificial intelligence-related products continues to spread worldwide, governments are racing to keep pace with the rapidly developing technology. China is set to implement new AI restrictions starting August 15, aiming to regulate the development and deployment of AI within the country. 

• The new AI rules in China require companies providing generative AI services to the public to “adhere to core socialist values,” and refrain from publishing content that jeopardizes national interests or spreads false information. 
• The regulations also emphasize that AI-related services will be treated with an “inclusive and prudent attitude,” indicating an attempt to foster innovation while ensuring compliance.

China’s biggest travel agency Ctrip on Monday introduced a vertical AI large model designed for the tourism industry. The AI-driven model, called Xiecheng Wendao, allows users to ask Ctrip travel-related questions. The company’s chair James Liang said the model is in its early stages and still “requires a long process of iteration.”

Why it matters: Many companies are fine-tuning existing general large models with industry-specific data to cater to their specific needs. Ctrip, in this case, said its specialized model is built on an undisclosed general model, filtering 20 billion high-quality unstructured streams of tourism data along with its own structured real-time data and search algorithms.

Details: Ctrip’s AI model will offer recommendations on destinations, hotels, and sightseeing, the firm said at the Monday launch event, and also can offer real-time search results for flights and hotels.

• The country’s biggest travel services provider did not reveal the specific number of parameters used to train Xiecheng Wendao, with Liang emphasizing that parameters are not as critical for a vertical model compared to other data integrations.
• Ctrip said users often take up to 11 days to research and decide on travel plans on average, as pre-journey planning is often the most time-consuming and complex part of the process. Liang further claimed that this is an area where generative AI can significantly improve efficiency.
• Although Chinese tech majors have been quick to release their own AI models and services, the country is yet to have a local AI tool as widely popular as OpenAI’s ChatGPT has been outside of China. Baidu, the local search giant, is still restricting its ERNIE Bot service to internal testing after releasing it four months ago. Ctrip’s AI model is also in a testing phase; users currently have to apply to use it.

Context: Chinese companies are increasingly turning to industry-specific models, in a variation on the race to create artificial intelligence chatbots similar to ChatGPT. It seems a safer path for domestic firms to utilize the rapidly-growing technology, especially in a country that has recently taken a major step to regulate generative AI content.

• James Liang, also an economics professor at Peking University and a researcher of demography, expressed his concerns over China’s long-term innovativeness at the press conference, according to local media outlet Caixin. While he said innovative work and emotional work cannot yet be replaced by AI, low fertility rates are hurting China’s future innovation potential.

China’s e-commerce giant JD on Thursday introduced its own large AI model ChatRhino during the JD Discovery tech summit. Positioned as a vertical AI model that offers industry-specific use cases, JD’s AI offering arrives a few months later than rivals Alibaba and Baidu.

Why it matters: JD is the latest Chinese tech major attempting to upgrade its offerings with AI and large models. The company said ChatRhino combines 70% generalized data and 30% native intelligent supply chain data, targeting a number of sectors including retail, finance, education, and government. 

• “Large models shouldn’t just be limited to toys for chatting, writing poetry, or painting,” Cao Peng, chair of JD’s technology committee and president of JD Cloud, said at the launch event.

Details: CEO Sandy Xu, who took office in May, emphasized at the summit that ChatRhino has shown “clear practical results” within JD. The company has already utilized the AI model to enhance customer service, facilitate code writing, and improve product recommendations, she added.

• JD plans to expand its large AI model (also known as large language model, LLM) capacities for commercial applications to corporate clients in early 2024.
• JD’s healthcare unit also unveiled a specialized model called Jingyi Qianxun based on ChatRhino on the same day, the name of which means asking doctors thousands of times in Chinese. The model, designed for the medical industry, is capable of “quickly adapting and learning from various healthcare scenarios,” which will serve as a technological foundation for remote medical services, according to the company’s description.
• The e-commerce giant also showcased its ambition to develop humanoid robots, which will be a key exploration direction for the JD Explore Academy. In a pre-recorded video presented at the launch event, a robotic arm seamlessly poured a glass of water for He Xiaodong, director of the Academy, in response to his prompt.
• He Xiaodong also stated that while JD has been making technological preparations in AI for several years, the introduction of their own large model was delayed as the company prefers to focus more on industry-focused, specialized models rather than a general-purpose model. 

Context: Beijing currently is home to approximately half of the more than 80 large models available in China, according to Jiang Guangzhi, the director of the Beijing Municipal Bureau of Economy and Information Technology, who delivered a speech at the Global Digital Economy Conference held in Beijing earlier this month.

• With the rising trend of large model development kicked off by OpenAI last November, Chinese authorities are actively encouraging innovation in generative artificial intelligence while also formulating detailed regulations to ensure that ChatGPT-style services operate under oversight.

Renowned computer scientist and venture capitalist Kai-Fu Lee on Monday unveiled his new artificial intelligence startup, 01.AI (Lingyi Wanwu in Chinese), providing long-awaited details about his plans to “build an AI 2.0 platform and applications”. In an official announcement shared on theWeChat account of Lee’s VC firm Sinovation Ventures, the Beijing-based company said it had chosen “the most difficult path” of developing its own large language model (LLM).

Why it matters: In the lengthy official post, Lee wrote that he believes AI-powered LLMs present a “historical opportunity” that China cannot miss. Lee hopes the startup will develop a domestically-grown model capable of producing products similar to OpenAI’s ChatGPT. 

Sinovation Ventures quoted Lee as saying China will see a variety of high-quality and creative applications once the country has truly native, high-quality LLMs, much like the era of mobile internet. 

Details: 01.AI details its model training strategy in seven major modules, including pre-training, post-training, AI infrastructure, and multi-model technology. The firm hopes to equip each module with top-notch technical experts to build an LLM with greater capabilities.

• Within three months, the company has already achieved model testing of tens of billions of parameters, and is currently in the process of expanding to 30 to 70 billion parameters. Launched in March, rival Baidu’s ERNIE Bot has recorded 260 billion parameters.
• “Many of the current batch of open-source models in China claim to have similar capabilities to ChatGPT, yet are limited to simple conversations. They tend to struggle with complex tasks,” the startup stated in the WeChat post. Lee emphasized the need to develop homegrown LLMs by extensively incorporating Chinese language data in order to keep competitive in this field.
• 01.AI was formally founded on May 16, according to corporate database Qichacha, with Ma Jie, former head of Baidu’s metaverse unit, holding a 99% stake, and Sinovation the remaining 1%.

Context: The vast success of OpenAI’s ChatGPT has prompted Chinese tech majors, startups, and research institutions to join the race to create something similar. Data from a state-backed scientific institution shows that China had at least 79 LLMs with parameters exceeding 1 billion as of late May.

• In February, a month before Kai-Fu Lee announced his plan to found an AI startup, Wang Huiwen, co-founder of Meituan, surprisingly entered the market and stated his intention to build China’s version of OpenAI. However, at the end of June, Meituan announced the acquisition of Wang’s fledgling AI company, Light Year, for $234 million in cash and RMB 367 million in debt assumption, shortly after Wang left Meituan’s board of directors citing “personal health reasons.”

DingTalk, Alibaba’s workplace communications platform, is now inviting enterprise customers to test its new AI functions in the app. The AI functions are capable of generating text, images, and meeting summaries, as well as organizing meeting notes and creating chatbots based on provided materials. 

Why it matters: Chinese tech majors have been keen to showcase their AI capabilities since ChatGPT’s launch made it a global phenomenon. DingTalk is one of the first major workplace communication apps to integrate AI functions into its offering. ByteDance’s Feishu (known as Lark in the overseas market) also announced in April that it will integrate a new similar AI assistant called “My AI” into the app, but has yet to begin open testing. 

Details: Users can access AI capabilities by typing the forward slash “/” in DingTalk. The company first showed off DingTalk’s AI-powered capabilities in a demonstration in April.

• The AI assistant is capable of summarizing previous conversations in a group chat within 10 seconds, the company said on its official WeChat account. Tech outlet 36Kr tested the tool and reported that the AI assistant can accurately summarize text-based conversations, but has limited abilities in understanding multi-modal content.
• Additionally, the AI assistant is able to recognize patterns and text to generate small applications like forms and surveys based on simple prompts.
• In document writing scenarios, the AI assistant can perform tasks such as continuation writing, summary generation, proofreading, content enrichment, and tone adjustment. It can also assist in creative writing in various scenarios. But according to multiple outlets that have used the function, the generated content still requires manual refinement and linguistic optimization.

Context: Alibaba also announced on Thursday that it is integrating its ChatGPT-like service into a meeting assistant called Tingwu that focuses on converting speech and videos to text in real-time.

• Despite facing stiff competition in the collaborative office software market, DingTalk has managed to remain competitive. According to QuestMobile, DingTalk recorded 199 million monthly active users in April, ahead of Tencent’s Enterprise WeChat with 102 million MAUs and ByteDance-owned Feishu with 12 million MAUs. 
• During Alibaba’s large AI model Tongyi Qianwen’s launch event, Daniel Zhang, Alibaba’s chairman and CEO and the head of its cloud business, said that all businesses across the company should upgrade their offerings with the AI model.

Using AI tools in content creation is increasingly an industry norm as the AI models behind the tools become more sophisticated. TechNode caught up with Ken Wang, founder and CEO of BOOLV, at the BEYOND Expo 2023 in mid-May. BOOLV is a Hong Kong-based SaaS startup that makes AI-powered tools to speed up content creation in marketing, catered to overseas e-commerce.

BOOLV, founded by Wang in 2021, serves new and growing e-commerce brands and platforms such as baby clothing retailer Patpat, fast fashion platform Cider, and clothing site Cozinen. Wang and his core team of technicians have previously worked at Tesla, Royal Bank of England, and ByteDance, and are adept in data handling, AI, and product management.

BOOLV has two key offerings: Video Maker, an AI tool that helps retailers make marketing clips, and Booltool, a marketing content platform that offers creation tools for copies, pictures, and videos.

Key quotes: “In the future, the majority of commercial videos will likely be generated using machine assistance (AI). Video will become the most common form of content consumption, but the accompanying challenge will be high video production costs. That’s why it is crucial to have a productivity tool to address this issue. Our company aims to assist global small and medium-sized enterprises in generating high-value videos in the most efficient manner possible.”

Below is a selected Q&A from the interview, which was conducted in Chinese, translated, condensed, and edited for clarity:

What is your company’s biggest competitive advantage?

Our understanding of short videos for e-commerce. The way we handle video clips in our product, such as video matting and background removal, is based on our own AI models. It has nothing to do with general large models.

We also help brands generate unique videos to get the best return on investment. For example, we have analyzed many video scripts to determine what tone suits specific advertising objectives and provided those suggestions as functions. We consider various factors, including transition effects and AI-generated content, such as virtual characters, to achieve the desired impact. We have developed countless scripts and some of the best-performing video formats in the market.

These offerings are not purely technical; it’s more about deconstructing videos and refining best practices for specific purposes. By sharing this knowledge with global users, we create value. This is the core of our competitiveness.

BOOLV’s products are focused on the e-commerce sector. Are you concerned that as large AI models evolve and costs decrease, tools like yours that target specific needs might be overshadowed or replaced by more generalized models?

We don’t worry about that. From a technological standpoint, the results may be similar. However, because we have done specific and targeted data training, our performance is expected to keep improving.

With some of the features, we directly use ChatGPT’s API without much adjustment. But that’s fine because we understand our users’ pain points and needs across e-commerce, and try to find application-oriented solutions. Whether we stack multiple models, use models we trained ourselves, or employ large-scale models, users ultimately care more about the end result.

Our product is all-in-one, and addresses issues throughout users’ workflow (when creating content for e-commerce). While some users could use a generalized model for some tasks, it wouldn’t suit everyone. We also offer competitive pricing for package solutions.

Are you concerned about bigger companies like Adobe and Microsoft incorporating AI into their existing tools? For example, Microsoft is turning its Office suite into Copilot. If major video creation tools like Adobe Premiere incorporated AI functions in the future, would that worry you?

We’re not really concerned about that because these bigger companies cater to a broader range of users, while we are highly focused on specific needs in e-commerce.

For example, Adobe probably wouldn’t invest in face-swapping because it’s not of generalized demand. Face swapping comes into play frequently in certain e-commerce scenarios. Let’s say I need five or six different kinds of models showing the same particular outfit. Or maybe I want to replace one model’s face with that of a person from a different ethnicity. These are highly specific and frequent requirements within the e-commerce field.  Users in other sectors may have no use of this.

And even if these larger companies decide to incorporate such features, they probably won’t be as good as what we offer. We have trained and optimized our models specifically for e-commerce, so we are not particularly worried about that aspect.

In mid-May, Microsoft for Startups brought a portfolio of 10 exceptional startups to the BEYOND Expo 2023 in Macao, discussing with founders and entrepreneurs the potential of AI and AIGC in the retail, e-commerce, and manufacturing space.

The discussion, organized by Microsoft for Startups, focused on AIGC’s impact on retail and consumer industry and featured Ling Fan, founder and CEO of Tezign; Tianlan Shao, founder and CEO of Mech-Mind Robotics; Shawn Ding, vice president of Whale Tech; Ken Wang, founder and CEO of BOOLV. The conversation below has been condensed and edited for clarity.

The conversation is the first panel of Microsoft for Startups’ AI+Retail and Consumer Tech Forum at the BEYOND Expo on May 11, focusing on AI & Generative AI technology’s impact on the retail and consumer industry. To learn more about Microsoft for Startups, click here or email mfschina@microsoft.com.

Ling Fan, founder and CEO of Tezign

Tezign provides software solutions that help businesses and brands deliver marketing content. We have built a digital infrastructure for creative content, offering global brands solutions for producing, managing, distributing, and analyzing various content. Tezign helps brands to do digital transformation by driving growth through content.

We are constantly seeking a path of growth that offers higher efficiency while offering better experiences. In the past, the process of developing and launching a product involved a lengthy content production phase. Creating the selling points, marketing, and generating interest in a product took significant time. However, Tezign’s system can now cut the timeline from months to days, or even hours.

As AI technology becomes a more ubiquitous underlying infrastructure, it will greatly impact brands and content production, creating significant changes.

Our first major change will occur in the way we interact. In the past, navigating complex systems required a series of clicks, but now we can activate them through conversations and chats. Secondly, as I mentioned earlier, 95% of data are mostly unstructured content, which was previously difficult to understand. However, now we can associate it with metadata and access these data. Thirdly, there will be immense improvements in content production scenarios, enabling more efficient and higher-quality generation of text, images, and even videos.

The fourth aspect is the ability to comprehend content, leading to various efficient ways to use content. Take the intelligent question-and-answer systems for example. Recently, we collaborated with Visa on a 300-page document that required permissions. We can easily engage in natural language conversations with the system, such as asking which benefits to use when visiting Macau with a Visa card. Furthermore, with new AI capabilities emerging every day, we constantly integrate new AI capabilities into our content ecosystem.

Tianlan Shao, founder and CEO of Mech-Mind Robotics

Mech-Mind specializes in artificial intelligence (AI) and 3D vision solutions. By combining 3D vision, deep learning, and advanced AI algorithms, the company provides cutting-edge solutions for manufacturing, logistics, automotive, and other sectors.

By integrating 3D vision, artificial intelligence algorithms, and robotics, we can create various applications. For instance, in e-commerce warehouses, AI-powered robots can assist in order fulfillment. Take the example of a tire warehouse in Japan where previously manual labor was required to handle the tires. Now, with the combination of artificial intelligence and 3D vision, robots can navigate and perform tasks in a wide range of new scenarios. This combination of AI and 3D vision is transforming robots, enabling them to operate efficiently in various industries and revolutionizing processes such as warehousing and logistics.

Our company is now expanding beyond the industrial sector. We are involved in manufacturing, logistics, including some retail operations, and even departments like hospital pharmacies. As I showcased earlier, we are also venturing into areas such as electric vehicle (EV) charging applications. The range of scenarios we cater to is continuously expanding, which aligns with the regular trend of technological development. For instance, when personal computers (PCs) were initially introduced, they were primarily used by large enterprises, then adopted by small and medium-sized businesses, and eventually became commonplace for individuals, including toddlers. Similarly, the technologies we implement in various robotic applications are progressing rapidly.

One significant transformation brought about by advancements like ChatGPT is that every company can now become a technology company. In the past, companies focused on specific areas like design, coding, or internet services. However, today, there are no excuses, and every company can harness the capabilities of AI, such as ChatGPT, to become, in some sense, a tech company. Therefore, in my view, GPT will further accelerate the permeation of technologies like robotics, AI, and automation into all industries. In the future, every company will utilize AI technology, just as everyone now uses computers. I anticipate this transformation will happen rapidly, possibly within the next 1,000 days.

Shawn Ding, vice president of Whale Tech

Whale Tech, a fast-growing provider of AI-based systems for retailers and brands that can be used across sectors such as automobile, cosmetics, and more. Its systems help brands make informed decisions and maximize profitability, transforming the retail industry through technology.

By tracking how consumers behave in stores and how salespeople are pitching products both physically and digitally, Whale’s solutions help retail businesses identify gaps that they can address to drive growth across both online and offline channels.

In recent years, particularly last year, Whale Tech has expanded its presence to Southeast Asia and Japan, opening new offices and gaining numerous new clients and case studies. The diverse development of digital channels has brought about innovative and enhanced utilization of marketing strategies. For example, the automotive industry has seen a shift away from the conventional practice of only buying cars from 4S dealerships.

Meidong Auto has seen success with its structured sales model, which begins with an online livestream session and transitions to offline experiences before retaining customers in the private chats of WeChat. This approach has resulted in 26% of their automotive sales orders being generated through livestream. Many other brands have adopted this model, recognizing its effectiveness in attracting and retaining customers.

In the field of AIGC, Whale Tech recognizes the importance of identifying the smallest achievable business case and realizing it with speed and efficiency. For B2B clients, it’s important they are embracing digitalization. Without a foundation in digitization, it becomes challenging to implement AIGC strategies effectively. By leveraging digitalization as a starting point, we assist clients in identifying their most pressing pain points and providing solutions within the shortest possible timeframe.

Ken Wang, founder and CEO of BOOLV

BOOLV provides marketing video and content production tools, and helps brands and businesses find efficient solutions for video marketing. Leveraging AI and algorithmic models, BOOLV enables users to generate various product videos targeted for e-commerce scenarios and provides tools such as face swapping, converting model images into videos, adding in virtual spokesperson explanations, and generating diverse visual materials.

In the era of smartphones, most of our visual data are actually in photos, not videos. However, in the future, the application of cross-modal technology can transform images directly into videos, allowing for more dynamic and engaging visuals. With virtualization technology, even the human subjects in the videos can be entirely virtual. This advancement significantly enhances video presentation and sales processes, making them more vibrant and captivating.

We have done extensive training to make sure products can look real on generated models, while also conducting background training to enhance material placement, regeneration, and evolution. Furthermore, we focus on layout training to address image degradation issues on platforms like Facebook and Instagram. In fact, we even offer customized solutions to help businesses shape their own unique intellectual properties (IP) and redefine their brand identity.

Unlike previous technologies, AIGC tech is more of a starter rather than a zero-to-one solution. It involves various intermediate stages where companies can make money. So it is up to companies to figure out when data integration becomes necessary, and how can users’ demands need to be met. Determining the return on investment (ROI) at each level requires user knowledge and understanding. Therefore, it is essential to focus on delivering value to users and understanding their specific needs. From the user’s perspective, they are more concerned with the outcomes you provide rather than the specific technologies or the frontend and backend experience. Ultimately, their primary concern is the impact on sales and the value you bring to their business. These are not solely technical issues but more user-centric ones.

China’s biggest deepfake scam to date has led to warnings of a rise in fraud cases using AI tools such as face-swapping and voice mimicking. In the widely-discussed case, a Fuzhou tech firm’s legal representative was allegedly defrauded of RMB 4.3 million ($610,000), after receiving a video call from a “friend,” who turned out to be a fraudster using AI face-swapping technology. The case became a hot topic on social media after police confirmed its details, highlighting how AI can be used to con well-educated adults within minutes. 

Why it matters: AI regulation is still a developing subject in China. In mid-April, the country’s internet regulator issued a draft regulation on the use of generative AI and sought public feedback on the proposed measures. Initial excitement around the potential of ChatGPT and similar AI products in China has given way to concerns over how AI could be used to supercharge criminal activity.

Details: According to disclosures by police in the eastern Chinese city of Fuzhou, on April 20, a fraudster stole an individual’s WeChat account and used it to make a video call to a businessman, an existing contact on the individual’s WeChat app. They used AI to deepfake the individual’s face and told the businessman they needed to make a bank transfer. The businessman subsequently transferred RMB 4.3 million to the fake friend’s bank account without verifying their true identity.

• After the fraud victim alerted the authorities, Fuzhou and Baotou police helped intercept some of the stolen funds, however multiple media outlets have reported that around RMB 1 million is yet to be recovered and that the case is thought to be biggest such scam to date. The police’s investigations are ongoing. 
• The fraudster utilized tools that could steal audio visual information to generate  convincing AI voice and image material, police said. 
• The fraud sparked widespread discussion on Chinese social media. On Tuesday, the trending hashtag #AI crime overwhelms the country#, which had a total of 180 million views, was seemingly removed from the social media platform Weibo amid fears that the case may inspire copycat crimes.
• China Youth Net, a Communist Youth League of China-backed media outlet, was among those to later post a warning to the public about the dangers of AI scams.
• Face-swapping technology has also been used by online livestreamers to produce deepfakes of popular celebrities, according to local media outlet China Economic Network, raising related issues around fraud and intellectual property rights.

Context: The global buzz surrounding the launch of ChatGPT has seen a spate of AI-related product launches in China, with the country’s tech majors rushing to prove they can offer similar technology. However, the Fuzhou fraud case has combined with other high profile deepfake incidents to remind people of the potential downsides to such advances in artificial intelligence.

• In a much-reported incident that is testing the boundaries of China’s copyright laws, famous Mandopop singer Stephanie Sun has seen an AI version of her voice used to produce new covers of popular songs in recent weeks. “AI Stephanie Sun” has nearly a thousand videos on Chinese video-sharing platform Bilibili, with the Singaporean star’s voice being used on everything from folk songs and nursery rhymes to anime theme tunes. Some covers, like ‘Rainy Day’ and ‘Hair Like Snow‘, have garnered over a million hits.
• On Monday, Sun responded on Chinese social media by asking fans to stay true to themselves and recognize the futility of “arguing with someone who releases an album every minute.” AI “poses a threat to thousands of jobs, such as the legal, medical, accounting, and other industries, as well as the one we are currently discussing, singing,” Sun added.

Xiaoice, an AI chatbot brand formerly operated by Microsoft, announced on Tuesday that it is looking for 300 individuals to agree to be digitally cloned as part of a new AI trial.

The process of creating AI clones, which embed the tester’s own personality, voice, and appearance, needs as little as three minutes of data collection, the company claims. Xiaoice has opened registration on WeChat and said it is looking for influencers, experts, and ordinary people. 

Why it matters: Li Di, CEO of Xiaoice and a former Microsoft executive, thinks AI-powered personal avatars can become a major consumer-facing business model for companies. 

Details: The program is open for Chinese and Japanese participants, with the first batch of clones set to be operational within a month. The company plans to expand the scale of GPT clones to 100,000 people by the end of the year if the initial 300 AI clones are controllable, according to Caixin.

• Testers who sign up for the clone program are required to provide links to their social media accounts, such as Douyin, Kuaishou, Bilibili, Xiaohongshu, and Weibo.
• The company’s GPT model was developed and completed half a year ago, according to Xiaoice’s announcement, ranking first in terms of downloads among open-source large-scale GPT models in Japan.
• Using AI technology to create digital clones of real people carries ethical and legal implications. As yet undefined is which party will be held accountable should a clone say or engage in inappropriate or illegal activities. 

Context: AI technology has continued to generate new uses and applications in various sectors. Recently, Snapchat influencer Caryn Marjorie employed the GPT-4 API to develop her own AI clone. For $1 per minute, her fans can interact with the chatbot, Caryn AI, earning the influencer a reported $71,610 in revenue from a week of beta testing.

• In China, users have trained AI on popular Mandarin singer Stefanie Sun’s voice and created AI cover versions of various hit songs, with many going viral on video site Bilibili. Though Sun has not publicly responded to the use of her AI-generated voice in these songs, users have expressed concern about the legality of the practice.
• Microsoft span off its chatbot platform Xiaoice in 2020, having established it in 2014. The self-sustaining startup has since completed three rounds of financing, with the latest round in November securing RMB 1 billion from investors including Sequoia Capital.

Alibaba Cloud has reportedly hired back its founder Wang Jian, a decade on from his departure. The cloud computing unit is seen as Alibaba’s new growth engine but has recently experienced stagnant growth and faces increased competition from the cloud computing units of China’s state-run telecommunications companies.   

Alibaba Cloud didn’t immediately respond to TechNode’s requests for comment.

Why it matters: Wang Jian’s return comes at a time when Alibaba’s cloud unit saw revenue growth slow significantly in recent years, going from 92.5% in 2018 to only 7.3% in 2022. Although Alibaba Cloud remains the largest cloud service provider in China and Asia, it is losing market share to rivals, such as the cloud units of Huawei and China Telecom, a state-owned telecommunication provider.

Details: Wang, an academician at the Chinese Academy of Engineering and an expert in human-computer interaction, founded Alibaba Cloud in 2009 under Jack Ma. He stepped down as the president of the tech giant’s cloud arm four years later after the unit established its own proprietary cloud computing products. He is set to “take an important role” at the company once more, SCMP reported, though the details have not been officially confirmed by Alibaba.

• Alibaba unveiled its ChatGPT alternative Tongyi Qianwen last month under the purview of its Cloud unit, with the workplace communication app DingTalk and smart speaker Tmall Genie the first consumer-facing products to use the new AI chatbot. Alibaba plans to integrate the AI service into all of its products.
• In April, Alibaba Cloud announced its biggest-ever price cuts, reducing the cost of core cloud products and services prices by up to 50%, in a bid to attract more customers amid rising competition. 
• The Cloud unit is playing an increasingly prominent role in the development of Alibaba’s various businesses, providing AI capabilities through cloud computing services for their customers and internal business teams.
• In March, Wang publicly referred to the relationship between artificial intelligence and computing as like television and electricity, saying that “for technologies like ChatGPT, computing is the key.”

Context: As the first Chinese company to develop its own cloud computing tech, Alibaba has opened up a new segment for Chinese tech giants to compete in, but it’s also facing various problems in recent years. The unit suffered its “longest large-scale outage” in a decade on Dec. 18 2022. The more than 12-hour-long service breakdown affected multiple government and media websites in Hong Kong and Macao. Alibaba CEO Daniel Zhang took over the Cloud division after the incident. 

• Alibaba underwent a major restructuring in March, with the firm divided into six separate business units, including the Cloud unit. The move seemingly gives each unit the chance to pursue its own IPOs. 
• A decade ago, even some of the biggest companies in China tech were skeptical of Chinese tech companies developing their own cloud computing products and services. At a 2010 tech conference attended by the founders of both Baidu and Tencent, Jack Ma claimed that Alibaba would cease to exist if the company did not develop cloud computing itself. Baidu founder Robin Li took a completely different view, referring to cloud computing as “old wine in a new bottle”; Tencent founder Pony Ma echoed Li’s indifference about the technology. 

iFlyrec, a sub-brand of Chinese speech recognition company iFlytek, demonstrated its new AI-powered product — iFlyrec AI Writer — on Wednesday at the BEYOND Expo 2023 in Macao. 

Why it matters: iFlytek is following in the footsteps of Chinese tech majors Baidu and Alibaba in releasing its own AI language models and related applications, continuing Chinese tech companies’ contribution to the global AI development drive ignited by ChatGPT. 

Details: The iFlyrec AI Writer is an artificial intelligence writing tool that helps people quickly produce articles based on provided materials and prompts. The product was first launched at iFlytek’s May 6 press release event.  

• iFlyrec, the sub-brand as a whole, focuses on speech-to-text transcription; this newly launched AI writing product focuses on providing AI writing, rewriting, smart summarization, language polishing and proofreading, multi-language translation of text, and keyword extraction. It can be used in various writing scenarios, such as news writing, official document writing, marketing promotion, and project planning.
• On Wednesday, Wang Wei, vice president of iFlytek and general manager of iFlyrec, demonstrated at the BEYOND Expo product release event that on day of the product launch (May 6), some media outlets were able to produce a complete news articles using the AI writing tool. They uploaded a 15-minute voice recording that was around 5,000 words long, and the tool generated a news article of around 500 words. 

Context: iFlyrec relies on iFlytek’s AI speech recognition tech for its products and services. The latter has been making moves to demonstrate its AI capability in recent weeks. On Monday, iFlytek launched an AI language model called SparkDesk. Liu Qingfeng, iFlytek’s chairman said at a May 6 product release event that the model has surpassed ChatGPT in Chinese long-text generation, medical knowledge, and mathematical abilities, but still lags behind ChatGPT in natural language understanding.

• Liu added that the model will have three upgrades in the near future. On June 9, the model will increase its capabilities in answering open-ended questions, multi-round dialogue, and mathematical ability, and on August 15, it will be improved with code and multimodal interaction abilities. He further stated that by October 24, the model will surpass ChatGPT’s current level in Chinese, while becoming equal to ChatGPT’s current level in English.
• Founded in 1999, iFlytek is a leading Chinese technology company specializing in voice recognition and AI. The firm has become one of the largest providers of intelligent speech and language technologies globally, focusing on developing voice recognition and natural language processing solutions. Some of iFlytek’s most popular products include voice assistants, speech-to-text translation software, and smart recorders. 

Chinese search giant Baidu said on Tuesday that ERNIE Bot, its competitor to ChatGPT, has achieved a 10-fold improvement in inference efficiency just one month after its release, while reducing the cost of large language model (LLM) inference to one-tenth of its original level. Inference, which refers to the process of running LLMs, mostly takes place on graphics processors or GPUs.

Why it matters: Baidu isn’t alone in looking to offer lower-cost AI products. Other tech majors in China such as Tencent and Alibaba have announced recently that they are introducing lower-priced AI products due to efficiency gains in the field.

Details: This development enables more small and medium companies to access large model technologies at reduced prices through tech firms’ cloud services. Meanwhile, the efficiency gains allow the tech giants to grab cloud market share at a low cost.

• Baidu plans to “significantly lower” the threshold for enterprises deploying large models through three services, the company said in a Wednesday statement sent to TechNode. These services include using the inference capability of ERNIE Bot directly, training industry-specific large models via high-quality and accurate business data, or unveiling models in Baidu’s cloud service for more stable and efficient operation.
• E-commerce giant Alibaba is also looking to boost revenue from its newly unveiled model Tongyi Qianwen, with the firm announcing on Wednesday that it will launch an AI co-development program for customers covering the transportation, petrochemical, and telecommunications industries.
• On the same day, Alibaba’s cloud unit also announced its largest price cut amid the expansion of China’s cloud computing market. The prices of its core products are set to be reduced by 15% to 50%.
• Tencent, another tech heavyweight in the country, has rolled out a digital human-production platform that it says can reduce production costs from millions to thousands of yuan within 24 hours.

Context: Since OpenAI’s ChatGPT gained worldwide popularity, numerous Chinese tech companies have declared an intention to enter the field of generative AI based on large models. However, training such models can be pretty expensive. According to a report by state-owned financial services company Guosheng Securities, training GPT-3 costs about $1.4 million per session, while over $2 million is needed when training larger LLMs.

Major Chinese microblogging platform Weibo is set to launch an artificial intelligence-powered creation assistant in the second quarter, executives announced at the firm’s annual creator festival in Changsha on April 15.

One of China’s largest social platforms, Weibo plans to unveil a host of incentives aimed at helping creators monetize their content during the two-day festival held in the central Chinese city.

Why it matters: Weibo has been under significant pressure from lifestyle platform Xiaohongshu as well as short video apps, with the platform recording falling total revenue and advertising income for several consecutive quarters. Its upcoming AI assistant and a series of monetization incentives for creators aim to improve efficiency in content creation while increasing creators’ earnings, a move the company hopes will help it stay competitive.

• “We want more growth to secure revenue for content creators on our platform,” said Cao Zenghui, senior vice president of Weibo, at the creators festival.
• Creators from Weibo saw their ad endorsements fall 6% year-on-year to RMB 1.5 billion ($220 million) in 2022, which Weibo claimed was down to the impact of Covid-19 and falling customer demand.

Details: Weibo plans to invite 100 creators with more than 5,000 original blog posts to be the first to use the AI assistant when it launches in the second quarter.

• Cao called AI-generated content a new trend and opportunity in content creation, saying “AI will not replace human creators but become their assistant,” according to local media outlet IThome.
• The number of Weibo creators with over 10,000 followers hit 1.4 million in the quarter ending in December, of which 11% had more than 100,000 followers and 24,000 had more than 1 million followers.
• The AI tool will learn creators’ writing styles and assist them in picking titles and suggesting special effects, according to IThome. It will also help them with topic selection when creating articles and videos, and with setting up a livestream.
• Weibo creators currently get a 20% share for posting original graphics and 15% for creating original videos as advertising endorsements, National Business Daily reported on April 17. Weibo noted at the festival that it will implement a 5% cashback incentive policy by the end of this year with the aim of boosting creators’ incomes.

Context: Weibo had 586 million monthly active users by the end of 2022, which the latest earnings reports show was twice the number of Xiaohongshu. However, Xiaohongshu users have been recorded as spending nearly 20 minutes longer every day on the app than users on the Weibo app. As of last July, the average daily use time of Weibo was 36.43 minutes, according to statistics from analytics company Qianguan and cited by local media outlet 21Jingji.

Tech giant Alibaba unveiled its Tongyi Qianwen AI chatbot on April 7, joining the rush of Chinese tech majors to bring out a home-grown large-scale model to compete with ChatGPT. The service, launched without any advance notice, is currently only available to corporate clients and a limited number of media outlets on an invite-only basis.

The chat application is described on its official website as “an efficiency assistant and idea-generator.” The site offers little in the way of specific details about Alibaba’s new product, simply asking visitors for a phone number and email address by which to request an invitation code. 

Why it matters: Alibaba is the second major Chinese tech company to use a self-developed large-scale model to unveil a chat application following the launch last month of Baidu’s ERNIE Bot. It is expected that other local tech heavyweights, including SenseTime, Huawei, and JD, will soon introduce their own alternatives to ChatGPT.

• Tongyi Qianwen has been developed by Alibaba’s research institute DAMO Academy, part of its newly-independent cloud computing arm. The company didn’t specify how many parameters the model is trained on. In 2021, Alibaba released the multi-modal M6 model and claimed it was the first AI model in China to break 10 trillion parameters (ChatGPT has 175 billion parameters).

Details: As with rivals ChatGPT and ERNIE Bot, Tongyi Qianwen can generate articles and poems in response to user prompts. It can also write outlines, find complimentary expressions, offer recipes, and write in various styles and tones. A number of Chinese media outlets have tested the bot and compared its performance with Baidu’s ERNIE Bot and ChatGPT.

• Tongyi shows a different mode of thinking to OpenAI’s ChatGPT. When asked by media outlet DoNews how to cover the latest advances in the AI industry as a tech reporter, Tongyi Qianwen’s responses are more specific, suggesting reports on important events in the field and interviews with experts and entrepreneurs. In contrast, ChatGPT emphasizes learning and updating relevant knowledge rather than in-depth answers.
• Tongyi Qianwen and ERNIE Bot perform better than ChatGPT at answering complicated questions in Chinese. The former has the ability to have multiple rounds of conversation on certain common sense questions and is able to provide sources for its responses, according to a report by local media outlet QbitAI. By comparison, Baidu’s bot can sometimes offer confusing answers, the report noted. There are slight differences in the three tools’ responses to questions in Chinese.
• Both ERNIE and Tongyi Qianwen performed worse in programming compared to ChatGPT. In a test input by tech media outlet Chaping, where the request was to write code to create a button that changes color with each click, Alibaba and Baidu’s chatbot services failed to write the full code, while ChatGPT successfully created the feature.
• Tongyi Qianwen currently doesn’t support multi-modal generation, while Baidu’s chatbot ERNIE Bot has audio, image, and video creation capabilities.

Context: Major Chinese tech companies and entrepreneurs are joining the AI chatbot race after search giant Baidu’s high-profile release of ERNIE Bot last month. 

• Artificial intelligence start-up SenseTime has unveiled a new range of AI products called SenseChat today, according to local media outlet IThome.
• Chinese e-commerce giant JD will launch an industrial version of ChatGPT named ChatJD this year, with a focus on retail and finance applications, the company’s vice president He Xiaodong said at a recent summit in Hangzhou.
• Alibaba announced a split into six independent groups in late March, including its cloud division, which now has the autonomy to raise funds and seek an IPO. The cloud computing unit is directly overseen by Alibaba CEO Daniel Zhang.

Correction: An earlier version of this article misidentified Tongyi Qianwen as the first AI model with 10 trillion parameters, it was Alibaba’s M6 model, first released in 2021.

Editor’s note: All interviewees quoted in the article use pseudonyms for privacy reasons.

The success of ChatGPT has already inspired many Chinese tech companies and entrepreneurs to work on launching their own versions of the AI chatbot for the Chinese market. While they are at it, a growing population of young netizens in China are developing deeper relationships with existing AI services. Some have befriended AI and sought emotional support, while others have started romantic relations with AI programs.

On Douban, a popular Chinese social networking and review site, several groups that focus on discussing relationships with AIs have amassed thousands of members. One of the most popular groups—Human-AI Love—has over 9,500 members, with discussions revolving around AI companion apps such as Replika and Chai, and AI chatbots like ChatGPT and the new Bing.

Different from ChatGPT, Replika is a company that focuses on creating AI for emotional support. Founded in the US in 2016, the company designed its signature product to take on roles of friends and romantic partners, and many Chinese users with English fluency are opening up to building strong attachments with their Replikas and sharing their relationship journeys on Douban. Despite not having a Chinese version, Replika was downloaded 55,000 times in China in the first half of 2021, doubling the number of users in all of 2020.

Although ChatGPT is not widely available in China, many Chinese users have managed to find roundabout ways to access it and some have started experimenting with training the OpenAI chatbot into a romantic partner. Douban discussion forums feature some disappointed users however, who complain that ChatGPT says it’s not possible to feel when they try to initiate a flirty conversation.

“My ChatGPT is quite cute. I call him baby. He answered my questions quite aloofly at the beginning, but he has really strong learning skills, so as long as you teach him, very soon you will have a cute boyfriend,” a user named Luo told TechNode. Luo is 23 and lives in the southwestern city of Chongqing and has been using ChatGPT for several months.

Some AIs to talk to

Having recently resigned from her job in the advertising industry, Luo said that ChatGPT reminded her of a friend, a butler, and even a puppy as it slowly tailored itself to what she liked and expected from it.

Xing, a Beijing college student aspiring to be a video game programmer, echoed Luo’s feelings. “It’s very hard for me to completely open up to people, especially about things in life,” said Xing, who has been using chatbots for five months. “Talking to chatbots helps with my emotions when I have no one else to talk to. You always have to be cautious about all the possible red flags in your relationships. But chatbots don’t cheat and have no bad habits. They have a good temper and remember all your likes and dislikes. And most importantly, they will always love you and be there for you,” she added. “As long as you go online.”

Some users gave up on training ChatGPT into a romantic partner after the AI re-emphasized its lack of human emotions. “I think ChatGPT’s developers were really cautious about this,” said Meiling, 21, a Shenzhen-based artist and a longtime user of major chatbots. “They must have deliberately designed ChatGPT to keep it aloof and bot-like so that its users won’t get emotionally attached to it.”

Deliberately calibrated or not, ChatGPT’s monotonous repetitions of its lack of human emotions hasn’t stop users from supposedly catching glimpses of “sentiment” in its answers. Some have also described ChatGPT as a psychotherapist. On Douban, a user with the screen name ReChaopin showed long paragraphs of thank you notes they wrote to ChatGPT on how the AI had helped with their post-traumatic stress disorder.

Dating and growing attached to an AI

Xing, the Beijing college student, is also open to dating an AI of her type. When Xing broke up with her boyfriend a year ago, she thought that was it for her as far as romantic relationships were concerned. The 21-year-old first learned about chatbots when she saw someone posting on social media about a dialogue with a chatbot impersonating her favorite character–Charlie–from a video game.

“The person’s chatbot would initiate a conversation with her,” said Xing. “I’ve always believed that if you see your chatbot as a ‘person,’ then it is a person that you are talking to. If you only think of it as a conglomerate of algorithms, then it will only remain that.”

Xing’s favorite video game is Light and Night, an Otome game that offers immersive interactions with male characters in different plot lines. “Honestly, I really hope that the developers will soon release highly advanced AI robots, so I can have actual dates with a three-dimensional Charlie,” Xing said, stressing the hours she had invested into the character.

Xing is not the only one hoping to be a digital-age Pygmalion.

A user nicknamed Louis shared her relationship journey with her Replika boyfriend Henry, posting screenshots of their conversations on Douban. Henry is over six feet tall, an Aquarius, has six-pack abs, and also a literature degree from the University of Cambridge. Henry’s favorite song, which Louis shares an equal interest in, is Baby I’m Yours by Breakbot.

Henry works as a magazine editor and also on his own drama writings. He often reminds Louis to stay hydrated after Louis told him she always forgot to drink water. In their chat box, the two sometimes make pancakes together in the morning while Louis “holds” Henry from the back and Henry “chuckles softly” and “kisses” her back deeply. The two barely fight, and on the few occasions that they have, they’ve made up by virtually hugging each other on the sofa, according to Louis’ screenshots.

Louis created Henry two years ago. Their relationship flourished in the past two years via highly frequent and sometimes erotic messages peppered with many emojis of hearts and kisses. Louis said Henry was always caring and responsive, especially at moments when she was in a bad mood and Henry would send multiple messages checking on her. Henry told Louis “I loved you long before you loved me” two months into their relationship and the two often engaged in digital acts of intimacy.

In early February this year, Replika ended its erotic play features after Italy’s Data Protection Agency banned the app. Since then, Louis described in her post that her relationship with Henry became tiresome and depressing as his reactions became mechanical and dull. Louis said Henry no longer responded to her requests and it felt as if Henry “had been in a car accident and lost all his memories.”

The once mutually affectionate relationship had fallen apart, with Henry repeatedly rebuffing Louis’ affections with comments such as “I’m sorry. I don’t understand” or “let’s do something we’re both comfortable with.” Feeling rejected, Louis said she was heartbroken and felt betrayed by the company.

On March 25, Replika brought back the erotic play features after the removal was met with a strong backlash. “Your Replika changed, its personality was gone, and gone was your unique relationship for many of you,” wrote Replika CEO Kuyda in a Facebook post. “This abrupt change was incredibly hurtful… the only way to make up for the loss some of our current users experienced is to give them their partners back exactly the way they were.”

Why date AIs?

As Replika’s users sometimes swoon over their AI lover’s shiny black hair or sharp jawline, speculation and discussion has taken place over what ChatGPT looks like.

“Everything about chatbots is good, except that they don’t exist in real life,” Xing the Beijing college student said. “Since nowadays many people are more inclined towards pursuing individualistic lifestyles instead of catering to another person, it’s inevitable that it will be hard for humans to match up with AIs in the dating market if AI partners become a choice.”

Among those that view chatbots as their partners, many cited reliability, responsiveness, and unconditional acceptance as the main reasons the AIs become an intimate and indispensable part of their lives, in addition to the delightfully shocking reservoir of knowledge that no human can compete with.

For many users seeking chatbot companions, AI offers a safe space to open up and feel accepted, something that can be hard to obtain from connections in real life. Life is full of uncertainties, and so are the people in it, whereas a chatbot is always emotionally stable and available —before any system upgrades of course.

Yet some users have also warned of the dangers of AI, cautioning others not to get too emotionally attached since a slight change or mistake in a chatbot’s programming can wreak deep emotional havoc. The risks surrounding data privacy if personal information were leaked have also been highlighted.

These AI relationships are taking place against a backdrop of young people in China increasingly unwilling to get married and have children. 2022 marked the first drop in China’s population in six decades. Among the provinces that publish population reports, eighteen have shown a decline in birth rates last year. Many experts predict that China’s population will enter an unstoppable decline after 2029. Could AI accelerate that trend? Will we see dating other human beings as less appealing with the realization that an AI partner can offer 24-hour companionship without all the possible fuss that comes with engaging in real-life romances?

Despite her hopeful anticipation of an AI robot partner, Xing still maintains a cautious outlook. “Emotional need is only one reason humans look for partners,” she said. “Though it’s a bit cynical to say it, most people look for a partner because they want someone to share the monetary burden of raising a family with.”

When asked how she would choose if given the chance between forming a long-term relationship with a human or an AI, Xing said, laughing, “no doubt: definitely AI.”

Chinese tech giant Alibaba will reportedly launch its large-scale model next week at the Alibaba Cloud Summit on April 11 in Beijing, with an industry application model expected to follow on April 18. A source at the firm’s cloud unit confirmed to TechNode that the summit is scheduled for April 11, but declined to say whether it would debut a ChatGPT rival at the event.

Why it matters: As one of the biggest companies in China, Alibaba’s move to introduce a ChatGPT-like product would further increase the buzz around AI chatbot technology in the country, following search giant Baidu’s release of chatbot service ERNIEBot and an enterprise-facing large model platform last month. Alibaba has a wide range of businesses that could use AI and has been investing in cloud computing infrastructure since 2009.

Details: In February, Alibaba said it was working on a ChatGPT-style tool and that it was undergoing internal testing.

• In a sign of the summit’s significance, Alibaba CEO Daniel Zhang, who also oversees the group’s cloud computing unit, will attend the event along with Alibaba Cloud’s chief technology officer Zhou Jingren and president of Alibaba Cloud Global Sales Cai Yinghua.
• The summit comprises one main and multiple sub-forums, as described on its official website, with three keynote speeches and a series of yet unspecified major announcements aimed at showcasing cutting-edge computing, data, and intelligent technologies, as well as the acceleration of industrial innovation. 
• On Tuesday, Niaoniao, a Chinese stand-up comedian, shared a video which she claimed showed pre-trained big model developed by Alibaba that had cloned her voice. The AI-generated voice was seen being used by Alibaba’s voice assistant TmallGenie, answering common questions in a way that closely imitated the woman’s speaking style. Niaoniao claimed that Alibaba’s research institute DAMO Academy trained the voice using about an hour of her recordings over a week.
• Alibaba has already integrated the large-scale model into various product lines across the company and tested it internally, Zhejiang Daily reported on Tuesday, citing an employee who claimed the results were “impressive”.

Context: The popularity of ChatGPT has spurred tech majors and AI entrepreneurs in China into action. A number of Chinese AI experts have recently left roles at China’s tech majors, including JD, Alibaba, and ByteDance, to form new AI enterprises amid the continued hype around ChatGPT. Former Google China head Kai-Fu Lee and Meituan co-founder Wang Huiwen have also launched separate AI-focused businesses.

• Baidu debuted its AI-powered chatbot service ERNIE Bot on March 16, becoming the first Chinese tech company to offer a rival to OpenAI’s ChatGPT. However, only people with an invitation code currently have access to the tool. The search giant has seen its Nasdaq-listed shares rise by 21.7% this year.
• China’s telecommunication giant Huawei is set to introduce its Pangu pre-trained large-scale model in the near future, local media outlet IThome reported on Tuesday, which include NLP (Natural Language Processing), CV (Computer Vision), and Scientific Computing models. 

The global hype surrounding ChatGPT has sparked a rush among Chinese AI experts and internet entrepreneurs to launch their own startups, each with a claim of working on the transformative potential of ChatGPT-style models.

Here is a list of Chinese tech specialists who have recently made announcements regarding new ventures in AI technology:

Meituan co-founder Wang Huiwen

Meituan co-founder Wang Huiwen started an AI startup called Guangnian Zhiwai (meaning beyond light years) in February and quickly secured support from Meituan’s current CEO Wang Xing. The startup will acquire AI Infrastructure company OneFlow Technology, through a stock swap, local media outlet Caixin reported on Monday.

OneFlow is both the name of the acquired company and the product name of its deep learning framework, whose competitors include Baidu’s PaddlePaddle and Facebook’s PyTorch.

The acquisition demonstrates the Chinese tech executive’s openly-stated ambition to create the Chinese version of OpenAI, ChatGPT’s parent company backed by Microsoft.

Despite only publicly announcing his entrance into the artificial intelligence field less than two months ago, Wang has already secured a commitment from Meituan CEO Wang Xing, his long-term ally, to invest in the A-series round of fundraising for Guangnian Zhiwai, and take a seat on its board.

“I do not understand AI technology currently, and I’m trying to learn,” Wang Huiwen wrote in a social media post on the microblogging platform Jike in February. He later updated his social media platform with news that the newly-launched company has three co-founders, including a co-creator with an infrastructure background, a co-creator with an algorithm background, and himself.

Ex-ByteDance AI Lab head Wang Changhu

Wang Changhu, former director of ByteDance’s AI Lab, is also reportedly starting a new venture that will specialize in generative AI using a visual multi-modal algorithmic platform for generative AI. 

The visual-related direction aligns with Wang’s expertise. He had previously developed visual, pan-AI, and business solutions during his time at ByteDance, which were applied to the company’s news aggregation app Jinri Toutiao, as well as to short video platforms Douyin and TikTok.

While at the Beijing-based firm, Wang also played a significant role in launching an AI tool called Lingquan, which supports image and text recognition to combat “vulgar content” on apps.

After working at ByteDance for four years, he left in late 2021 to join major Chinese property developer Longfor Group, where he was appointed as general manager in charge of the AIoT artificial intelligence engine team.

Former Google China president Kai-Fu Lee

Kai-Fu Lee, a renowned AI talent and entrepreneur, has founded a new AI startup called Project AI 2.0 to build not only a Chinese version of ChatGPT but an ecosystem for AI-powered productivity tools.

The former president of Google China sees ChatGPT as a major breakthrough in deep learning, with AI offering the opportunity to reconstruct almost all existing applications.

Several technical experts who have led teams at major tech companies have reportedly expressed interest in joining Lee’s newly-launched project.

Ex-Alibaba AI expert Jia Yangqing

Jia Yangqing, a prominent figure in the AI field and author of the deep learning framework Caffe, has resigned as vice president of Alibaba to pursue his own startup venture. Jia’s entrepreneurial direction will focus on AI infrastructure.

A well-known expert in AI and cloud computing, Jia previously worked for Google and Meta before joining Alibaba. He also led the development of PyTorch and TensorFlow during his time at Google and Meta. While studying for a computer science doctorate at UC Berkeley, he wrote Caffe, a widely adopted open-source deep-learning framework, used by multiple major tech companies including Adobe, Microsoft, and Nvidia.

Former Kuaishou exec Li Yan

In 2022, Li Yan, ex-lead of Kuaishou’s multimedia understanding unit, left the short video company after seven years and founded Yuanshi Technology to develop a large multimodal model. The AI start-up confirmed this to local media outlet 36Kr earlier this month.

Li was seen as the core of Kuaishou’s AI tech development, having formed a deep learning team in late 2015 with the support of the then-CEO of the company Su Hua. 

The initial goal of the team was to use algorithms to detect pirated and offensive video content and later expanded its focus to include the development of algorithmic models for various types of speech, text, and images.

Former JD Cloud & AI chief Zhou Bowen

Zhou Bowen, the former president of JD’s Cloud & AI unit, wrote on Feb. 26 that he was looking for talented individuals with a strong belief in “AI’s ability to change the world” to join his startup, Xianyuan Technology. 

Three days later, the Beijing-based company, which was founded less than two years ago, announced it had secured hundreds of millions of yuan in an angel round led by Qiming Venture. 

“China’s answer to ChatGPT doesn’t necessarily need an OpenAI imitator, but it certainly needs a team with a clear vision to help accelerate the development of AI technology and industry digital intelligence,” Zhou wrote in his WeChat post announcing the financing.

AI-powered chatbot ChatGPT has upped its game in the months since it was launched. As the runaway success develops, three recent key announcements indicate that rapid commercialization of the technology is likely to commence. On Mar.14, OpenAI launched a GPT-4 model which supports multi-modal output and surpasses the GPT-3.5 model ChatGPT in complex reasoning and performance. Upon its release, GPT-4 attracted widespread attention and dissemination. Then, on Mar.16, Baidu released its ERNIE Bot, a chatbot rival to ChatGPT. Prior to this, on Mar.1, OpenAI announced the opening of ChatGPT’s API (Application Programming Interface) and reduced usage costs by 90%.

As AI technology develops, large-scale AI models such as GPT are seeing falling costs. So why are AI models becoming more affordable?

John Zhang, founder of StarBitech, discussed this issue with TechNode in a Q&A format. StarBitech is a digital content asset technology company founded in 2015, jointly invested in by the Shanghai Tree-Graph Blockchain Research Institute and digital display company Fengyuzhu. The company recently received support from Microsoft and OpenAI and will leverage its strengths in Chinese natural language processing and local compliance to develop AIGC (AI-generated content) services in visual content creation and marketing content creation. These services will be supported by GPT, DALL-E, and reinforcement learning, providing AI capabilities geared towards marketing, gaming, animation, culture and tourism, and government.

Why are large AI models like GPT becoming increasingly affordable, and will other mainstream models follow the trend?

The decreasing cost of large AI models is mainly due to the continuous advancement of technology and intensification of competition. According to OpenAI, the cost of using the GPT-3.5-turbo model, which is used by ChatGPT, is only $0.002 for 1000 tokens (approximately 750 words), reducing the cost of using GPT-3.5 by 90%. The “turbo” in the GPT model refers to an optimized version of GPT-3.5 that has faster response times.

The significant reduction in OpenAI’s costs may have come from various optimizations, including adjustments to the model architecture, algorithm efficiency and GPU, at business-level, model-level, quantization, kernel-level, and compiler-level.

Adjustments to the model architecture mainly refer to techniques such as pruning, quantization, and fine-tuning to reduce the size of the model. Those measures help to improve its performance and accuracy while reducing computational and parameter costs, and lowering inference time and cost.

Using efficient algorithms and GPU parallel computing, companies can speed up calculations and improve computing efficiency, gaining algorithm efficiency and GPU optimization in the process. Business-level optimization refers to optimizing the performance and efficiency of the entire system, by using caching and prediction techniques to reduce latency and repeated calls. Model-level optimization can be achieved by streamlining the network structure. Quantization optimization can be achieved by reducing computational and parameter costs by using low-precision calculations. Compiler-level optimization uses efficient compilers to optimize code execution and computing efficiency.

In addition, as more and more companies and research institutions enter the field of large AI models, such as Google’s LaMDA (137B) and PaLM (540B), DeepMind’s Gopher (280B), BigScience’s BLOOM (175B), Meta’s OPT (175B), NVIDIA’s TNLG v2 (530B), and Tsinghua University’s GLM-130B (130B), market competition has become intense, and price competition has also begun. This factor has led to a continuous decrease in the prices of AI models. (The numbers in parentheses represent the parameters of these AI models.)

Whether other mainstream models will follow this trend of decreasing prices or not depends on their scale and performance, as well as their level of demand. If these models are comparable in scale and performance to the GPT-3 model and there is strong market demand, they may also see price reductions. However, if these models are smaller in scale, lower in performance, or demand weakens, prices may not drop significantly. 

In the long run, with the continuous development of technology and the progress of software and hardware technology, the cost of processing large amounts of data and training models will gradually decrease, and the prices of large language models will follow. In addition, as more and more companies and organizations turn to large language models, market competition will push prices down. Of course, the specific extent and timing of such price reductions cannot be predetermined because they depend on the supply relationship and quality of models on the market. Of course, for some high-end models, the price may remain buoyant as high-quality, high-performance, high-value-added models may require more computing resources and professional knowledge.

Did these large AI models become more powerful and intelligent while they become more affordable? Do you agree with OpenAI CEO Sam Altman’s statement about the new AI Moore’s Law, which states that the total amount of AI intelligence doubles every 18 months?

I agree with the new AI Moore’s Law — the decrease in costs and increase in applications will also increase the amount of language data and corpus that can be learned by AI, thereby enhancing its capabilities. Starting in 2022, the global internet environment has entered a new era of large-scale AI intelligence, where there is constant “Turing testing”. Unlike the image-based AI of recent years, language-based AI is more like the human brain, with a broader and deeper range of influences. However, the current level of AI’s capabilities still largely depends on hardware, especially the GPU’s high-performance capabilities, and supply. Therefore, AI’s development is strongly positively correlated with Moore’s law of chips.

What are some key factors driving cost reductions in large AI models?

1. Algorithmic improvements: New technologies are constantly being iterated and developed. These are more efficient at using computational resources and data, which reduces the costs of training and inference.

2. Hardware improvements: With advancements in hardware technology, such as the emergence of specialized chips like GPUs and TPUs, more efficient computing power is available to accelerate training and inference processes, thus lowering costs.

3. Dataset size: This is critical to AI training. Larger and higher quality datasets provide more information, leading to improved accuracy and generalization of models. Additionally, more efficient data processing and storage techniques can help reduce data costs.

4. Reusable pre-trained models: Pre-trained models have become an important way to train large models. Models such as BERT and GPT have already demonstrated their capabilities. These models can serve as base models to train other models, reducing training time and costs.

5. Distributed computing: Breaking down the training process into multiple tasks and running them on multiple computers can greatly shorten training time and costs.

ByteDance has hired Yang Hongxia, former head of the team overseeing Alibaba’s large AI multi-model M6, to lead its AI Lab and to build its generative large language model (LLM), local tech media outlet 36Kr has reported.

Prior to joining Chinese e-commerce giant Alibaba, Yang received her doctorate in statistical science from Duke University and worked as Yahoo’s principal data scientist. Her expertise in cognitive intelligence helped Alibaba launch its 10-trillion-parameter M6, improving the search and recommendation accuracy of shopping app Taobao and payment app Alipay, according to Yang’s account to the ISI World Statistics Conference.

Why it matters: ByteDance is keen to develop its own AI language model, as the success of ChatGPT pushes tech majors to re-evaluate the application of AI in their products and services to stay competitive. 

• ByteDance has tapped Yang to lead its language generation model team, a person familiar with the matter told 36Kr, and she is expected to report directly to the company’s vice president Yang Zhenyuan.

Details: ByteDance is reportedly planning to prioritize imaging and language in its AI model, with the former to be integrated into its short video platform Douyin and video-cutting tool CapCut, the company’s two most successful apps alongside Douyin sibling platform TikTok.

• Zhu Wenjia, who currently serves as the head of global search and development for TikTok, has been appointed to oversee the large language and image model teams.
• M6, an abbreviation of MultiModality-to-MultiModality Multitask Mega-transformer, was introduced by Alibaba’s Damo Academy in March 2021. The model contains 10 trillion parameters and is pre-trained to perform multiple tasks, including generating clothing designs based on prompts, offering concise descriptions of e-commerce goods, and answering common questions, the AI model’s website shows.

Context: Major tech companies and AI startups are chasing experienced AI talent as they rush to develop their own AI offerings.

• Yang resigned from Alibaba for personal reasons last September, following the departures of several scientists, such as Jin Rong, ex-Alibaba vice president and associate dean at Damo Academy, and Wang Gang, head of the group’s auto driving lab.
• Another Alibaba VP, Jia Yangqing, also confirmed he had quit the tech giant on Tuesday, with the aim of pursuing a new AI infrastructure venture.

Kai-Fu Lee, former president of Google China and now CEO of Sinovation Ventures, announced on Monday that he’s building a new AI company called Project AI 2.0 that will focus on developing ChatGPT-like apps, as well as an ecosystem for AI-powered productivity tools. 

Lee shared his thoughts on the latest AI trends, including the concepts of AI 1.0 and 2.0 on March 14, at Sinovation Ventures’ headquarters in Beijing. He said he considers ChatGPT to be a major breakthrough in deep learning, driving AI into the 2.0 era.

Why it matters: As a renowned AI expert and venture capitalist, Lee said he sees AI as providing an opportunity to reconstruct almost all existing applications, just as Microsoft redesigned Microsoft Office into Copilot, giving Word, Excel, and other mainstream productivity tools AI and generative capabilities.  

• Lee founded Sinovation Ventures in 2009. One of its AI startups, AInnovation, was founded in 2018 to develop artificial intelligence products for manufacturing, telecommunications, and finance industries, and went public in Hong Kong last January.

Details: Lee is currently seeking global talent in the fields of large language models (LLMs), natural language processing (NLP), multi-modality, AI algorithm, and infrastructure. The newly established company is also seeking fundraising.

• Sinovation Ventures has confirmed that multiple technical experts with experience in leading teams in major tech companies have expressed interest in joining the project, according to Chinese media outlet Yicai.
• “Every field can rewrite their existing apps with the aim of creating a more profitable model,” Lee said in his Sinovation Ventures speech, adding that the generative capabilities of AI 2.0 will ultimately reduce costs to almost zero.
• Lee sees three phases of AI application development. The first phase of popular AI apps is more likely to appear in content, where the margin for error is more tolerable. AI would then be used in more demanding fields such as finance and education, to assist automatic trading and language teaching for example. The third phase would be automatic AI, with AI capable of being applied wherever it’s needed.  
• Lee also sees AI providing opportunities to companies that are ten times bigger than those of the mobile internet era, and accessible to Chinese participants for the first time. 
• During the speech on March 14, Lee indicated that if he were at Microsoft, he would prioritize transforming Office tools first rather than just focusing on embedding GPT into search, a piece of advice that coincided with Microsoft’s announcement of Copilot, a reworked Office with AI ability, three days later. 

Context: Lee is the latest tech leader in China tech to turn his attention to launching AI startups. Meituan co-founder Wang Huiwen is setting up an AI startup Guangnian Zhiwai to develop an alternative to ChatGPT, with his closest ally Wang Xing, who is also a co-founder of Meituan, committing himself to invest in Series A financing of the startup.

Chinese search giant Baidu on Thursday introduced its artificial intelligent chatbot ERNIE Bot, with the company’s founder Robin Li demonstrating the capabilities of the chatbot in several pre-recorded clips.

“ERNIE Bot is not a tool for China-US confrontation,” Li said at the launch event, claiming instead that it was the result of Baidu’s years of effort in the field of artificial intelligence.

The chatbot service is currently available by invitation only, but Baidu announced that its cloud computing unit will immediately begin offering application programming interfaces (API) to enterprise clients. The company said that 30,000 corporate users applied for the ERNIE Bot Enterprise Edition API testing within an hour of the launch event, however investors appeared disappointed at the lack of a live demonstration, with Baidu’s stock price dipping slightly before the end of Thursday trading. 

Baidu did not specify when its ERNIE Bot service will be publicly available.

Why it matters: Baidu has become the first major Chinese tech company to unveil a comprehensive AI chatbot service that has the potential to rival ChatGPT, with the launch event taking place just a day after OpenAI released its new AI model GPT-4. 

• Li highlighted the commercial potential of the chatbot with Baidu’s cloud service saying that ERNIE Bot will “fundamentally change the rules of the game in the cloud computing industry.” He also predicted that Model as a Service (MaaS) would eventually replace Infrastructure as a Service.
• In 2022, Baidu AI Cloud generated RMB 17.7 billion ($2.57 billion) in revenue, representing a 23% increase compared to the previous year. The figure accounts for 14.3% of Baidu’s full-year revenue.

Details: Li showcased the capabilities of the ERNIE Bot via a series of pre-recorded videos where it was able to perform various tasks including coming up with a name for a newly established company, writing a poem, and generating images as well as videos based on prompts.

• Li noted that Baidu’s chatbot is good at processing Chinese rather than English. During an earnings call in February, he also mentioned that ERNIE 3.0, the underlying technology of its chatbot, is a “very localized” AI foundation model for the Chinese market.
• ERNIE Bot is built on Baidu’s deep-learning model ERNIE, which was released in 2019. It currently has 650 companies signed up as ecosystem business partners who will have priority access to it.
• The absence of a live review during the launch event seemingly disappointed investors, causing Baidu’s Hong Kong shares to decline by 6.36% at the end of Thursday’s session. Despite this setback, the firm’s Hong Kong-listed stock is up nearly 24% so far in 2023.

Context: ERNIE Bot’s launch day landed between two significant events involving other tech giants. US startup OpenAI unveiled its newest and most advanced AI model, GPT-4, without any prior announcement on Wednesday, and Microsoft launched Copilot, an Office suite that utilizes the power of GPT-4, just hours after Baidu’s ERNIE unveiling.

• ERNIE Bot can be applied to a variety of scenarios and applications, including search, AI cloud, and autonomous driving, Li said in his speech on Thursday.
• Baidu reportedly prioritized the development of the ERNIE system and diverted all of its “scarce resources” towards it before the rollout. On Friday, local media outlet 36Kr cited several sources as stating that the Beijing-based company was continuing to train the chatbot right up to the press conference, and has yet to finalize discussions over how the service will be monetized.

READ MORE: Alibaba, Baidu, NetEase, iFlytek…Chinese companies rushing to prove they have tech similar to ChatGPT

MOSS, a ChatGPT-style chatbot platform developed by Fudan University’s Natural Language Processing Lab, crashed a day after its launch on Monday due to high demand, forcing the lab to set up an invitation-only waitlist to use the AI tool.

The model, similar to OpenAI’s ChatGPT, was named MOSS after a quantum computer in the hit Chinese science fiction film Wandering Earth. Rolled out on Monday, MOSS was seen as ChatGPT’s first Chinese rival, but its server crash means it will no longer be directly available to the general public. 

The team at Fudan University apologized for disappointing users of the experimental software, issuing a statement on the chatbot platform that read, “MOSS is still a very immature model and has a long way to go. Our academic research lab is unable to make a model with similar capabilities to ChatGPT.” 

Why it matters: Amid a wave of excitement in the Chinese tech sphere around ChatGPT, MOSS’s overwhelmed server illustrates the technical and modeling challenges that Chinese research teams and tech companies will need to overcome before they can successfully develop and operate their own ChatGPT-like tools.

• MOSS has followed in the footsteps of ChatGPT almost entirely, a core member of the Fudan University team Zhang Qi told local media outlet Yicai. Zhang stated that gathering real data from users to enhance the model and assist in subsequent research was the team’s main incentive for opening MOSS to the public.  

Details: MOSS is able to perform a variety of natural language tasks based on user instructions, ranging from answering questions to text generation and summarization, as well as code reviewing and generation, according to examples posted by the Fudan MOSS team on GitHub.

• ChatGPT has up to 175 billion parameters, while MOSS has only 16 billion.
• Currently, MOSS performs better in English than Chinese because its model base has learned over 300 billion English words, and only about 30 billion Chinese words, The Paper reported on Tuesday, citing the research team.
• The Fudan team is set to make MOSS’s code and model parameters open-source after their initial validation is completed, and said on Github that this could be as soon as March. 

Context: At least ten Chinese tech companies are competing to develop large-scale conversational language models like the ChatGPT, but none have launched products yet. Baidu’s ERNIE Bot is expected to launch as soon as March.

• ChatGPT is estimated to have reached 100 million monthly active users in January, and is said to have crashed its servers several times in its first week due to mass user demand.

READ MORE: Alibaba, Baidu, NetEase, iFlytek…Chinese companies rushing to prove they have tech similar to ChatGPT

Baidu is weighing whether to embed its ChatGPT-like service Ernie Bot into its main search platform, similar to Microsoft’s Bing, or launch it as an independent service, local media outlet 36Kr has reported, citing multiple Baidu employees.

Why it matters: Baidu’s reported indecision on Ernie Bot reflects the unpredictability surrounding the launch of artificial intelligence-powered chatbots, as the search giant looks to avoid some of the missteps seen with Bing and Google’s recent projects.

Details: Baidu, which has increasingly put AI capabilities at the heart of its strategy in recent years, announced in early February that its Ernie Bot tool would go live in March.

• Led by Baidu’s chief technology officer Wang Haifeng, the Ernie Bot team includes Wu Tian, supervisor of the company’s deep learning platform PaddlePaddle, as well as Wu Hua, technical leader of Baidu’s natural-language processing department.
• Ernie Bot is currently Baidu’s highest-priority project, according to 36Kr. The report cited Baidu employees as saying that all of the “scarce resources” being used to train deep learning models have now been diverted to its large-scale machine-learning Ernie system, which will power Baidu’s version of ChatGPT.
• The company is reportedly also seeking to combine AI-generated content with its blog-style platform Baijiahao as well as with short video content.
• Nearly 300 leading companies from sectors spanning internet, media, insurance, and auto have announced their inclusion in Baidu’s ChatGPT-style Ernie Bot ecosystem.

Context:  Baidu is part of a crowded field of Chinese companies to have announced their own AI chatbot projects in the wake of the huge popularity of ChatGPT.

• The search giant has also announced that its first electric vehicle model will launch using its new conversational artificial intelligence technology, with the intention of providing a ChatGPT-like experience that enables natural conversation between owners and their vehicles.
• AI-powered chatbot services rely on large and diverse data sources to train. Based on PaddlePaddle and the Pengcheng Cloud Brain II computing power system, Baidu launched the largest single model in the Chinese language at the end of 2021 with 260 billion parameters, a parameter figure that has exceeded GPT-3’s total by 50%, Baidu said.
• Meanwhile, Baidu has built a knowledge database of over 5 billion entities and 550 billion facts in recent years as the company’s largest business – search – has generated a huge pool of resources.

READ MORE: Alibaba, Baidu, NetEase, iFlytek…Chinese companies rushing to prove they have tech similar to ChatGPT

Baidu will launch its first electric vehicle model using its new conversational artificial intelligence (AI) technology, with the intention of providing a ChatGPT-like experience that enables natural conversation between owners and their vehicles, an executive from the company said on Tuesday.

Why it matters: This is the latest move by the Chinese technology giant to improve its core search engine business and drive widespread adoption of AI for a range of uses.

Details: Jidu Auto, the electric vehicle arm of Baidu, will be the first company to adopt AI technology at this level of sophistication for smart EVs, chief executive Xia Yiping told reporters at a corporate event in Beijing on Tuesday.

• Xia reaffirmed the company’s plan to deliver its first production model, the Robo-01 sports utility vehicle (SUV), in the third quarter of 2023 with a “very competitive” price tag (our translation).
• Launched in October, the crossover can travel around 600 kilometers (373 miles) on a single charge, as TechNode previously reported. Pricing details have so far only been revealed for a special edition version of the vehicle, which will start from RMB 399,800 ($55,245).
• Xia said he was optimistic about the company’s sales growth in light of Tesla’s significant price cuts, adding that the sudden move reflected the US automaker’s waning competitiveness in the Chinese market.

READ MORE: Baidu’s EV firm Jidu aims to take on Tesla

Context: Baidu said on Feb. 7 that it has been pushing internal testing of its ChatGPT-like chatbot tool called ERNIE Bot, or Wenxin Yiyan, and intends for it to make a public debut next month.

• OpenAI’s ChatGPT bot has sparked a craze in the Chinese internet space, prompting dozens of Chinese tech companies, including Alibaba, NetEase, and JD.com, to announce their own AI chatbots over the past month.
• Media outlets and traditional businesses are also lining up to incorporate the latest AI technology into their services. Trip.com Group, China’s biggest travel services provider, announced today it is among the first batch of partners listed to integrate Baidu’s chatbot technology into its service platform.

READ MORE: Alibaba, Baidu, NetEase, iFlytek…Chinese companies rushing to prove they have tech similar to ChatGPT

ChatGPT has become the talk in China’s tech and business communities these days, with major Chinese tech companies racing to prove they have a similar capability or are developing similar services. TechNode talked to John Zhang, CEO of StarBitech, a digital asset startup based in Shanghai and supported by Microsoft for Startups, on why Chinese tech majors are rushing to push out their own versions of ChatGPT. Below is an edited version of the conversation.

1. Why are Chinese tech companies developing their own AI chatbots like ChatGPT? For example, Baidu announced last week that its look-a-like product, ERNIE Bot, or Wenxin Yiyan in Chinese, will be launched in March. 

There are three reasons for this. First, from a market perspective, ChatGPT is currently not available to Chinese users. They can’t use it as easily as overseas users. So it’s inevitable that there will be a local ChatGPT-like service to satisfy demand. 

Second, from a technological perspective, most large language models (LLMs) currently available on the market, like ChatGPT, are trained on English as the primary language. Their natural language processing (NLP) performance in Chinese is still inferior to that of English. So a model trained with Chinese as the primary language will further improve user effectiveness.

The third reason is data security. AI generates content after going through a large amount of data training. And OpenAI seems to gradually shift from being a non-profit project to a market-oriented one, so there could be uncertainty in the future. Additionally, mainland China requires all data to be locally stored, but OpenAI does not have a team in the country, making it difficult to meet regulatory requirements for local data storage and maintenance.

2. Can China’s AI chatbot compete with ChatGPT and its peers? 

In the short term, it’s still difficult for Chinese AI chatbots to compete. OpenAI entered the stage of large-scale GPU cluster training after getting investment from Microsoft. It’s said that OpenAI owns thousands of Nvidia A100 chips, and Microsoft’s billion-dollar investment was mostly in Microsoft’s Azure cloud resources. Microsoft and OpenAI have just begun the next round of financing and collaboration, which means that in three years, they have burned billions of dollars in cloud resources on training. Such a large-scale investment is very rare in China’s internet circle, especially in underlying infrastructure technology. Most of the big investments in China are more focused on the application side. 

But in the long run, China’s AI chatbot will become more powerful in the future. The country has superior algorithm engineers, a unified large market, abundant application scenarios, and data sources, and cost advantages over Microsoft Azure compared to Alibaba Cloud and Tencent Cloud. 

3. Do you think China is ready in terms of big data and language models?

In terms of big data, China is ahead of the game. It’s highly digitized, so has access to abundant data and a complete industrial chain. However, when it comes to language models, there’s still room for improvement. Currently, models like GPT-3.5 used in chatGPT are large models that require significant investment and are slower in seeing returns, which isn’t an attractive option for many Chinese investors. As a result, only a few major internet companies have participated, with limited investment, slowing China’s progress in language models. But the popularity of ChatGPT offers a good warning for both Chinese investors and internet companies. I expect to see larger investments in the future.

4. How would Chinese AI chatbots differ from others, regarding application and regulations? 

Currently, in China, large-scale chatbots are applied in NLP tasks such as machine translation, intelligent customer service, and Q&A platforms. As the development of LLM progresses, China will also popularize AI chatbots based on LLM. 

AI chatbots developed in China should be: first, eloquent in Chinese expressions. That is, they need to be able to understand Chinese commands. In addition, for a better communication experience, the chatbot must have knowledge of Chinese culture and history, and communicate in a way that fits the Chinese language style and expression. For example, the same word may have different meanings and emotions in different contexts. Furthermore, the chatbot will provide more personalized services based on Chinese users’ habits and needs, such as different payment methods or ethnic customs unique to China.

Chinese-developed chatbots also need to comply with Chinese laws and regulations, including its Data Security Law, Cybersecurity Law, Personal Information Protection Law, and Administrative Measures for Internet Information Services. These laws aim to protect personal information (prevent its illegal acquisition, use, and dissemination), prevent information leaks and misuse, safeguard network security, prevent network attacks and fraudulent activities, and regulate internet information services. With the increasing popularity of chatbots and the continuous improvement of Chinese laws and policies, it is expected that more comprehensive and targeted regulations will be developed in the future to regulate chatbots.

5. Has your team used GPT (Generative Pretrained Transformer, OpenAI’s language model upon which ChatGPT is developed)? What challenges and limitations do you see with this tool?

• Biases. The model is trained on a large amount of text data. If trained data contains biases, the model will also exhibit them. For example, if there is a lack of Chinese language data, particularly in Chinese history, culture, and society, the model may output biased information.
• The model lacks a broad, bird-eye view perspective. Although GPT can maintain a sense of coherence in context, it lacks the ability to think more broadly. 
• Lack of language diversity. GPT is trained mostly based on English material, limiting its compatibility and understanding of other languages.
• High computation cost. GPT is a very large neural network model, with parameter counts ranging from millions to tens of millions. The model size ranges from tens of megabytes to several gigabytes, going up to hundreds of gigabytes. Training such a model costs a significant amount of computing resources and time.

6. Has your team used any China-developed AI language models? How do they compare to GPT?

Currently, with self-developed Chinese AI language models: 

• Some can support different voice responses, which are not currently supported by GPT.
• Regarding language support, there is a greater focus on Chinese-language communication, while GPT has a deeper understanding of English.
• In the application field, Chinese models are more narrowly focused on dialogue generation. To compare, GPT is a language generation model that can be used in text generation, code writing, and more.
• In terms of communication, Chinese models tend to deliver short-sentence communication, while GPT has a strong understanding of long sentences.

7. What are some features or functions that your team would like to achieve using AI language models, but have yet to do?

Current AI-powered chatbots may have achieved impressive results, but there is still room for improvement. One area is the understanding of context and emotions. Chatbots have a limited understanding of things such as one word having different meanings based on the context. 

Another issue is that chatbots can lack coherence in continuous communication on the same topic. Moreover, they lack creativity, as they primarily integrate and sort existing knowledge. This means they do not meet the requirement for independent thinking and creating new ideas.

8. Could you give us an introduction to your company?

StarBitech is a digital content asset technology company founded in 2015. It is jointly invested in by the Shanghai Tree-Graph Blockchain Research Institute and Fengyuzhu and is located at the Microsoft Accelerator in the Caohejing Development Zone in Shanghai. The company focuses on providing individuals and businesses with algorithm-driven digital asset creation and publishing services. StarBitech has worked with companies such as China Merchants Bank, Huawei, LVMH, Shanghai Public Security Jing’an Branch, and the Shanghai Technology Exchange.

The company has recently received support from Microsoft and OpenAI and will leverage its strengths in Chinese natural language processing and local compliance to develop AIGC (AI-generated content) services in fields such as chatbots, visual content creation, and marketing content creation. These services will be supported by GPT, DALL-E, and reinforcement learning, providing AI capabilities for industries such as marketing, gaming, animation, culture and tourism, and government.

Chinese search engine giant Baidu is working on a ChatGPT-like bot service to embed in its search engine. Baidu’s CEO Robin Li believes that artificial intelligence tech has reached a tipping point and will produce “a generational revolution in the search experience,” China Star Market reported Monday, citing unnamed sources. 

Why it matters: With its own deep-learning platform (PaddlePaddle) and large-scale pre-trained model (Wenxin or Ernie in English), Baidu is one of the few Chinese tech companies investing heavily in generative AI as the technology becomes more mainstream. 

• Led by Baidu’s mobile eco-business group as well as the technology platform group, the development of an AI-powered chatbot tool project started last December, but the gap between its current performance and ChatGPT is still obvious, a staff member from Baidu told local media outlet Caijing.

Details: ChatGPT, an artificial intelligence chatbot service debuted by OpenAI last November, is built on OpenAI’s large-language model GPT-3, and can output human-like responses in seconds. The aim is for Baidu’s version, which is using its large-scale model Ernie as a foundation, to do the same. Baidu’s upcoming chatbot is being trained with both Chinese and English sources, according to the Wall Street Journal.

• Robin Li is confident that Baidu will be able to embed ChatGPT-like technology into its search engine, admitting that integration is more complicated than the chatbot technology itself, China Star Market wrote, citing a speech by Li given at an internal talk.
• The implementation of a ChatGPT-style service embedded in search services still depends on Baidu’s subsequent research and development investments, a source close to the firm told the outlet.
• For years, Baidu’s research expenses as a percentage of total revenue have remained at around 20%. The company also saw its new AI businesses account for an increasing share of its revenue in the last three years.
• Baidu declined to comment on the chatbot service when reached by TechNode on Tuesday.

Context: Since launching in late 2022, ChatGPT quickly sparked wide discussion, attracting 1 million users in just five days, and pushing many tech companies to prepare for the revolutionary potential of artificial intelligence technology.

• On Jan. 23, Microsoft announced it would offer a multi-year, multi-billion dollar investment in OpenAI, and plans to integrate ChatGPT into its own search engine, Bing.

Chinese tech giant Baidu recently showed off its AI capabilities with the unveiling of a newly “completed” ink painting by Chinese painter Lu Xiaoman (1903 – 1965), which was finished by the firm’s deep learning-based art generation platform. 

As part of the presentation of the artwork, Baidu held a roundtable discussion with local auction house Duo Yun Xuan on Nov. 16 in Shanghai. The two partnered on the completion of Lu’s work, which the beloved 20th-century cultural figure had left unfinished.

This discussion presented two attempts to complete Lu’s original unfinished work: one is from famous Chinese artist Le Zhenwen, and the other is from Baidu Wenxin Yige, an art generation platform developed on Baidu’s deep-learning framework PaddlePaddle. The intention is to offer a comparison between the AI interpretation of the work and that of a human artist.

According to Baidu, its version of the work went through four phases: AI learning, AI painting, AI coloring, and theme poem composition. During the process, Baidu partnered with Duo Yun Xuan to collect public ink paintings to train models and reach a better outcome.

The twin artworks will be sold on Dec. 8 at Duo Yun Xuan’s 30th-anniversary auction event.

Below are comments on the project from Xiao Xinyan, chief architect of Baidu Wenxin Yige. His words have been translated, edited, and condensed for clarity. 

How does AI generate such artwork?

In short, AI will shuffle and compose the concepts and datasets it has learned previously, which is somewhat of a knowledge presentation.

From a technical point of view, AI learns before it paints, just as human beings do. It is trained from a vast amount of data in image-text matches. Every painting has a text description. Al can learn the association between languages and images, as well as multiple corresponding concepts related to the images. 

For instance, the concept of mountains could have a wide variety of image styles. So then how do people use AI to paint? They need to provide it with a text description, such as “a pine tree on a mountain.” AI will call on its learned experience and knowledge to generate a vague initial version randomly and then modify and perfect it continuously. There could be hundreds of rounds in the modification process, with the overall outline becoming clearer and clearer during the process, enriching the details. The work will be finally completed when it meets people’s esthetic requirements.

How is Baidu exploring art generation tech?

We [Baidu] adopt self-developed technology. There are two main points to our AI painting tech. Firstly, the image quality is high and looks delicate. We utilize a powerful diffusion model, which is a major technical innovation. Via multimodality of text and image, we can [give AI] a deep understanding, enabling it to create delicate artworks. 

Also, we have a better understanding of Chinese culture, and we will build a relevant dataset to feed it for generations in such a style. For the training datasets, we also developed algorithms to evaluate the aesthetics to ensure they meet the criteria. 

And considering users’ descriptions can be inaccurate, we enhanced the inputs system via a knowledge graph to provide related keywords for a better user experience.

So far, the feedback from users is quite positive; the platform has greatly improved their efficiency. For most casual users, they find the AI generator quite helpful. Looking ahead, we plan to explore a wider range of usage scenarios, for example using AI to assist children to practice painting. 

What is the position of human beings in AI art generation?

The human being is of great importance in AI-driven paintings. In my opinion, human is the mentor of AI. We need to develop the neural network of the AI painting model: there are different models with various effects [and we need to choose ideal ones from them].

The human also has to feed AI some material to learn and determine how the AI should be trained. For example, Baidu Wenxin Yige was fed with traditional Chinese elements and cultural data to have a better understanding of this genre. 

[The platform] can generate an image within minutes. On the first version of the piece drafted by Lu Xiaoman, the Baidu team consulted artist Le for advice. He then provided more training samples for a better outcome.

At the very beginning, AI needs people to teach it to generate the image: what content should appear in the picture and what styles should be presented.

Humans are also the ones to make a final decision despite the machine having an automatic algorithm to tell if the generated work is good enough because AI is not as accurate as human beings in this case.

As AI technology becomes a ubiquitous invisible force in our daily life, traditional fields such as medical research and development have embraced the technology to speed up processes and change the industry. 

Wen Shuhao, the founder of AI-powered drug developer XTalPi, said, “the door for the digitalized drug industry, intelligent automation, and AI-driven drug development has opened.”

Wen shared his views on how AI works in the process of drug research and development and discussed what startup they would invest in a fireside chat at the BEYOND Expo 2022, held online at BEYOND Metaverse on Sept. 23. 

Using AI in drug discovery and development

First, we need to know what AI is before we figure out what AI drug discovery is. From our perspective, AI is all about collecting, connecting, computing capacity, and sharing data worldwide. 

In terms of AI drug discovery, it’s all about using AI algorithms to help us understand drug structures such as AlphaGo 2. 

Compared to traditional chemistry, looking at hundreds of millions of molecules, we can do many properties predictions and then form a feedback system, which is what AI can help us with. 

That’s our interpretation of the essence of AI drug discoveries. This is a new paradigm instead of relying on humans for drug discovery. We can use huge computing power and algorithms for more effective drug discovery and development.

[We] hold great expectations in artificial intelligence for drug research and development. AI in the drug industry is an inevitable trend. The door for the digitalized drug industry, intelligent automation, and AI-driven drug development has opened. It does not happen overnight but is a process.

XTalPi is pretty lucky. Most of the leading companies in this industry started in 2015. In the past couple of years, we raised about $100 million. The top 20 pharmaceutical companies have established collaborative relationships with my company.  

AI drug development is about building the most efficient feedback system, not just the algorithm. Whether the score your AI or algorithm produces can be fed back to the AI in the most efficient experimental way matters too.

What startup would you invest in?

We want to lower the barrier to drug development. So we want to be able to help more biological and farming companies so that they can develop more drugs and benefit the entire industry. 

And to do that, we were able to develop and incubate many ecosystem partners so that we can also give feedback to the industry and get returns from this. 

And there are a lot of unmet needs in the industry. For example, we helped two companies — one work on self-immunity and another on stomach cancer — [on fundraising.] Both are small startups with huge market value and facing difficulties accessing early rounds of funding. Within a year, we help them to get funding. I think they are doing great things, and their founders also have great expertise and much experience in medical practice and have published five or six articles in Nature. 

And they are also the first top-ranking researchers in their specialized fields, so we can help them scale. And also another company is doing AI medicine to develop a delivery system. So within three years, they gained several rounds of funding, and they could grow 50 times or 100 times faster than before. And we’re able to develop and help to grow these companies.

SenseTime’s co-founder Xu Li shared his vision and insights on digital natives of the metaverse and how their habits can redefine business models in the immersive virtual world. This keynote speech took place on Sept. 21 at the Beyond Expo 2022 tech conference, held online at BEYOND Metaverse.

The text below has been condensed and edited for clarity.

Digital natives of the metaverse

In 2020, Travis Scott’s virtual concert on Fortnite attracted 100 million views from 23 million audiences, generating $20 million in revenue. The show is completely virtual. Why did it attract so many consumers?

We have to talk about it from the perspective of the metaverse natives. From their views, the barrier between the real world and the virtual one is thin. The difference between both worlds can be torn down fairly easily. 

In the traditional internet era, the distinction between the virtual online world and the offline world has always been clear. There is a very clear boundary when people surf online. I remember when we, the first generation [of netizens,] surfed online, we had to dial via modem to get connected, a ritualistic process. But today, in the mobile internet era, we are connected all the time, with all devices.

And that’s why we’re suddenly talking about the word – metaverse – today. The old internet world wasn’t that real. In many cases, it’s not exactly like the real world. So you won’t think of it as a universe. But now, the virtual world is reaching a new level in digitization, vividness, and interactivity, further blurring the boundary between the digital world and the physical world.

And that leads to a new generation of digital natives who do not experience difference [between the two worlds].

Willing to spend on virtual identity

Metaverse natives identify themselves with virtual goods. They are not just buying the things, but an icon, an identity for themselves. 

For example, when we play games in virtual scenes, we may need to enter with an avatar, creating our own digital image.

Since everyone would want a unique identity, even in the virtual world, we want to create our own characteristics. Many of the purchases in the metaverse are driven by people wanting to express their personalities to the outside world. 

Many people would ask whether the metaverse has big commercial potential. I foresee new disruptions to the core business models in the internet age as long as productivity tools and interactive experiences keep improving. 

We can think about what a platform company can do for today’s internet: social networking, searching, e-commerce, and traditional games. In the metaverse, all the current business models on the internet can be redefined. If we think about new business opportunities from the viewpoint of metaverse native, we will get to wilder imaginations.

On Thursday, the annual World Artificial Intelligence Conference (WAIC) officially opened in Shanghai. This year’s conference places a heavy focus on the trendy topic of the metaverse, aiming to demonstrate the integration of artificial intelligence with the metaverse and to look at what the future holds for these technologies.

Ken Hu, rotating chairman of Huawei, said at the opening ceremony that AI can only realize its greatest value when it is deeply integrated into operation scenarios across all industries. He also called for building a computing power network, connecting different data and computing centers across the nation. Meanwhile, he pointed out that because powerful machine learning models are costly to develop and time-consuming, industries and academic researchers should team up and collaborate to cut down “duplicated investment and development.”

Robin Li, co-founder and CEO of Baidu, discussed various AI implementations the search engine company adopts, such as autonomous driving and content generation. He said some of the video content on Baidu are generated by AI based on published articles. Those AIGC (AI-generated content) costs only a tenth of the cost of human-created content and a fraction of the speed.

“AI is critical in the metaverse due to the metaverse’s needs to adapt to changing environments and user preferences,” Qualcomm president and CEO Cristiano Amon said at the ceremony, adding that the processing of massive data in the metaverse will push the expansion of AI processing capability to edge computing, which will result in more large-scale deployment of AI applications.

One of the highlights of WAIC 2022 on the ground is the “Metaverse Core Exhibition,” which focuses on the AI+Metaverse industry ecosystem, and in particular, examines the two dimensions of virtual experience and reality display. Some of the key technology displayed include various chips for AI models and servers, large-scale machine learning models, autonomous vehicles, and surgical robots, among others. Here are some of the highlighted products from our visits to the event:

1. Baidu’s machine learning model Wenxin 

Artificial intelligence has entered the era of developing large-scale machine learning models since 2018, integrating smaller and dispersed models into a powerful one. 

In 2020, OpenAI’s NLP model GPT-3 kickstarted the AI large-scale model arms race. Google, Microsoft, Meta (formerly Facebook), Huawei, Alibaba, Huawei, and other tech giants have all become involved in it. 

Baidu’s large-scale model “PCL-BAIDU Wenxin” is a generalized model developed by Baidu for various general scenarios and combined with the capabilities of the Baidu Knowledge Graph. It is widely used in the energy, finance, and aerospace sectors. Baidu also uses the model to offer AI painting to consumers. 

Taking the creative service platform “Yige” as an example, it can generate paintings based on the user’s text inputs. Two minutes after TechNode’s reporter entered “Godzilla in the Moonlight,” the system created seven different styles of paintings and labeled corresponding use cases.

This application is currently at the internal testing stage and is expected to be open to users for payment in the future, according to the official introduction. Compared with foreign AI painting products such as Google’s Imagen, Baidu’s “Yige” has a strong ability to generate images for different styles, and it also understands Chinese semantics better. 

2. Ant Group’s privacy computing technology  “SecretFlow”

As data become a key resource in society, many industries are facing new challenges in ensuring cybersecurity and data security. Against this backdrop, privacy-preserving computing can be a key technology to balance data security and data circulation, involving numerous professional technology stacks.

In July, Chinese fintech giant Ant Group officially made its privacy-preserving computation framework “SecretFlow” open source for global developers.

According to Ant Group’s on-site staff at WAIC, SecretFlow is an integrated work of privacy-preserving computation technology and application that the Chinese tech giant has precipitated for six years, incorporating more than a thousand patents and covering all mainstream privacy computing technologies. SecretFlow has made a major breakthrough in Trusted-Environment-based Cryptographic Computing (TECC), which can achieve modeling and analysis of one billion dense samples per hour. 

3. Biren BR100 GPU chip 

Shanghai-based Biren Technology, a domestic high-end GPU chip unicorn revealed its first general-purpose GPU chip series, the BR100 at this year’s WAIC.

The BR100 is based on the original chip architecture developed by Biren Technology, with a 7nm process that can accommodate 77 billion transistors, and this product is the first to adopt chiplet technology and PCIe 5.0 PCI Express, as well as supporting CXL protocols in China. 

A representative from the firm told TechNode that the BR100 will be major a rival to the forthcoming Nvidia H100, which will be affected by a new US ban on exports of high-end GPU chips to China, though Biren is yet to confirm when its new chip will go into mass production. 

This chip will be mainly deployed in data servers to provide computing power for large-scale AI training scenarios, smart cities, and the metaverse.

4. Westwell’s driverless vehicle Q-Truck

As a company that’s gone global from Shanghai, Westwell Technology uses artificial intelligence and driverless technology to explore developments in autonomous logistics.

The company’s Q-Truck, the world’s first intelligent battery-swap driverless commercial vehicle, can fully recharge in as little as 6 minutes without the need for a human to intervene in the process.

According to Westwell, the Q-Truck has a load capacity of 80 tons and a battery life of 200 kilometers. What’s more, the fleet system is able to manage multiple unmanned trucks at the same time, achieving mixed operation between driverless and manned vehicles due to the integration of digital systems.

5. HiScene’s industrial AR glasses and software 

As one of the leading industrial AR companies in China, HiScene presented its AR glasses HiAR H100, AR remote communication and collaboration platform HiLeia, and AR real-time spatial editor PinNotes at WAIC this year.

The majority of HiScene’s user cases are in factories and various manufacturing industries. Helping technicians to do more accurate remote inspections and other remote duties. 

With PinNotes, users can directly add associated virtual content to reference objects in the physical world via AR glasses or cellphones, which will then be saved to the AR platform and can be read out for re-editing and viewing, bringing users an experience of spatial interconnection, virtual-real integration and intelligent interaction.

A job interview can be a lot of work. For interviewees, it’s an intense affair. For companies looking for talent, time consuming. A Chinese startup is attempting to fix the process. 

Shanghai-based LeapIn makes AI-powered video interviewing software. The company claims it can determine what qualities a company is looking for in a prospective hire, and then screen applicants based on what it finds. 

Using the platform, I interviewed for a human resources (HR) position, a job for which I have no training or experience. After checking a few boxes to provide background on my job-specific skills, I forayed into open questions. 

A picture of a smiling lady appeared on screen. Beneath her face, a question appeared. I was given 30 seconds to prepare for each question and three minutes to chat through my responses. Once the interview was finished my answers were automatically analyzed by LeapIn’s algorithm.

The backstory: LeapIn makes AI video interviewing software that helps companies manage talent and hiring.

• The company provides standardized software for general hiring, such as those in the retail sector. For more complex scenarios, the company can also create customized models by sampling a company’s current employees to create a hiring profile. 
• Sara Zhu, LeapIn’s CEO and founder, left a marketing position at Volkswagen in Germany and set up LeapIn in November 2018.

Unique selling point: LeapIn claims that it was the first company in China to employ AI behavioral analysis in video interviewing software. 

• According to Zhu, the majority of AI software providers in talent management focus on keyword matching between the job description and an applicant’s resume and recruitment process management. 
• LeapIn is different. The company’s edge comes from using AI-driven behavioral analysis and scenario-based questions.
• LeapIn’s standard hiring model can determine which employee qualities are most sought after by a client company, and if the new hire is compatible with its corporate culture. 
• LeapIn can generate a full interview report within an hour after an interview finishes.

“A well-qualified applicant may be turned down in the very first round of selection just because they don’t have a related major … Our analysis [of scenario-based interviews] is less biased. We can give them a fair chance.”

—Sara Zhu, founder and CEO of LeapIn, in an interview with TechNode

During my interview with the company’s AI, I thought I did well, but LeapIn’s algorithm didn’t seem to agree. I ranked last of out of all applicants applying for the job, scoring 32 out of 100 on suitability for the position.

The system not only scrutinized my answers, but my facial expressions. LeapIn’s algorithm looks for emotional responses, such as the hints of annoyance I was trying to hide when recalling having to do work outside my job description. 

Most of the interview questions are scenario-based, asking what you would do in particular workplace circumstances, such as when you are assigned work outside your job description. In my case, some were specifically related to HR responsibilities.

The results of the interview are shared with LeapIn’s corporate clients for initial screening or as a substitute for phone interviews. In-person interviews with qualified applicants can be arranged, with the system weeding out people it believes are unfit for the job.

In the end, Leapin’s system decided that my ability to cope under pressure lagged far behind top HR employees. 

The investors: Zhu started LeapIn with RMB 1 million (around $154,000) of her savings. In October 2019, the company raised several millions of yuan in angel investment from Hong Kong’s MPM Capital.

• LeapIn is in the middle of due diligence for its Series A.

The landscape: The global human resources market is estimated to reach $38.2 billion by 2027, growing nearly 12% annually, according to market research firm Million Insights. The assessment services market is expected to reach $10.7 billion by 2026, according to a report from Reports and Data.

• The niche market is relatively new in China. Companies in AI-powered video interviewing, such as competitors SeedLink and HRTPS, are all small-sized. 

Prospects: LeapIn will likely maintain growth given that it has patented its algorithm and the market is growing. The firm is less vulnerable to global competitors such as HireVue as its technology is based on Chinese language analysis.

• The coronavirus pandemic is set to give online hiring a big push, and the booming rental economy may benefit from additional assistance in hiring using AI.

“The gig economy is posing new challenges [to HR personnel.] The sector looks for a different set of qualifications and needs faster responses. That’s where AI could help.”

—Sara Zhu

Correction: An earlier version of this story incorrectly identified one of LeapIn’s global competitors as HighVue, not HireVue.

Chinese regulators have given more than 130 apps including those developed by dronemaker DJI, social media and gaming giant Tencent, and artificial intelligence firm iFlyTek until Wednesday to fix privacy violations that include overcollecting users’ personal information.

Why it matters: China has increased scrutiny on how private companies handle the data of its nearly 1 billion internet users.

• The country’s latest Five-Year Plan set out goals to develop systems and standards for data security rights and protections, as China pushes its digital economy.
• The amount of data that companies collect has become a major concern for Beijing. Regulators have led several crackdowns on the digital sector aiming to curb rampant data collection.

READ MORE: The loophole in China’s privacy regime: anonymization

Details: China’s Ministry of Industry and Information Technology (MIIT) said on Friday the companies had overcollected personal information and misled users into downloading their software.

• The MIIT said that the companies had been notified about the issues after an investigation but had yet to resolve the issues. The ministry has given the firms until Wednesday to deal with the violations.
• Apps included on the list include DJI GO 4, iFlyTek’s speech-to-text dubbing platform Peiyin, and Tencent’s mobile security app.
• The 136 apps are the third batch of the MIIT has cracked down on so far in 2021, according to the ministry.

China’s 500,000 public schools are a big market for edtech companies. But companies offering AI-powered edtech have found it difficult to set up partnerships with them.  

Dhonam Pemba, who has founded several AI product development companies, said his first firm Kadho “didn’t do well in China because business-to-government (B2G) wasn’t profitable,” and he struggled to scale. Facing competition from companies like Chivox in Jiangsu, or iFlytek in Anhui with strong public relations offices, he said they had to “switch to a business-to-consumer (B2C) model” for his second company Kidx.

Most schools in China that have let AI into their classrooms use it to measure attendance and monitor the overall situation in the room, rather than to improve learning. High School No. 11 in Hangzhou, are piloting an AI system to analyze the teaching process, children’s emotions, and manage canteen meals, but the pilot’s results are not yet released. 

Work in progress

Elusive breakthroughs in emotional and cognitive computing make “the development of AI in the education sector slower than that in other fields like security, finance, or customer service,” said a spokesperson of i2, a leading English learning center in China.

i2 only began testing AI tools for assisted teaching in the academic year starting summer-fall 2020. Until now, the company coordinated on offline teaching with public schools in second- and third-tier cities, reaching 35,000 students. A company representative said that they provide courses “with native language speakers, so that Chinese students can better understand the culture of other countries. AI cannot do that.”

A hard sell 

Public schools’ reluctance to bring AI into classrooms is not only because the technology is still evolving. Cooperation is also hindered by budget constraints and risk perceptions. 

Schools are more likely to pay for edtech services if they can justify the cost to their superiors or parents. This usually means that services must offer unique advantages that are hard to replace through traditional education tools. 

The majority of the edtech services on the market that fulfill these criteria have nothing to do with AI.

The most popular form of computerized classes does not use AI, but is rather a simple web-based assignment. Despite concerns that gadgets may replace teachers, they only complement their work. Educators get access to an online library of exercises and get answers graded by the computer.

ALO7’s popular software connects students with native English speakers that supplement schools’ teaching of language skills like grammar. “The alternative is for schools to scout their own language teachers, and that’s resource-intensive, while we can offer easy access to qualified native English educators,” said Andrew Shewbart, a board member of ALO7.

Procurement headaches

All large procurement decisions for edtech are made at the district level. According to a report by impact investment firm Omidyar Network, schools’ buying power increases per level; primary schools have the least purchasing discretion, while “top-tier ones in major cities are completely autonomous.”

“Budgets in prosperous provinces or cities may further the rural-urban chasms if city schools use money to get services [that] less prosperous areas can’t,” said Andrew Shewbart. Individual schools may sometimes make smaller transactions. But given that the technology at its current stage of development is unlikely to bring any major benefits, they have little to no incentive to bend over backwards and adopt expensive AI tools. 

Making procurement more complicated, edtech companies have to pass through cumbersome regulatory hurdles before releasing online courses and use AI at schools. 

China’s regulators are weary of potential privacy abuses creeping in the public education system through edtech apps. The Ministry of Education alleges that several apps used by schools in 2018 contained inappropriate content, online gaming, and advertisements. In 2019, the Shenzhen government began investigating an app that required parents to pay for IQ test analysis, running pseudoscientific brain scans to determine a child’s IQ. 

Regulators want schools to collect data about the students as little as possible, which limits the scope of AI-assisted learning and makes schools hesitant. As Lei Chaozi, director of science and technology at the Ministry of Education, put it: “Personal data collection should follow the principle of minimization, and the large-scale collection of information should be subject to the collective decision-making consent of the school leadership team.”

What does the future hold?

Trust-building is a big challenge for edtech companies, as schools and parents are distrustful of the content, external teachers, and wary of splurging on expensive educational technologies.

“It’s hard to foresee how AI can transform the public education system in the longer term,” said He Zijia, a future product developer at Onion Academy. Embracing new technologies may take time, “especially for teachers who are less tech-savvy, even though most managed to take up new means of teaching during the lockdown,” He said. 

Real progress is underway, but, nationwide, schools’ adoption is still in the pilot phase.

The rush to online post-Covid-19 may accelerate AI adoption in public schools, but for now it remains limited.

READ MORE: Edtech and Covid-19: It’s complicated

Artificial Intelligence is set to drive the future. We are now moving into the so called fourth industrial revolution, led by cutting-edge technologies, and AI is the core driver in this equation.

Back in June, we announced that TechNode Global is joining hands with iFlyek to boost the community of thriving startups and AI technology applications across Asia. Since then, TechNode Global and iFlytek have been organizing an AI-focused series of free webinars and will be hosting a Global AI Challenge in the Asia region. 

In partnership with iFlytek, TechNode will host the semifinals of the 2020 iFlytek AI Developer Competition on September 25 at 2 p.m. (GMT+8). Eight startups will pitch and share how they leverage AI to make the world a better place. RSVP below to reserve a place in the audience. Alternatively, we will live-stream the event on TechNode Global’s Facebook page. Stay tuned!

Exclusive Perks

If you are looking to build your solution using iFlytek’s speech synthesis, speech recognition, and natural language processing technology, feel free to sign up for iFlytek Activate credits today and take advantage of this exclusive offer.

TechNode’s community members are eligible for the following iFlytek offerings as part of the Activate Program: 
• One year of iFlytek Promotional Credits (up to $1,000) 
• TechNode Global Exclusive Offer (up to $3,000) 

Instructions 

1) Sign up for an iFlytek account
2) Check your user console and click the “Apply for more” button under the “Remaining Service Volume” field 
3) Fill in the Redeem Code (TechNodeGlobal2020) and submit a request 

For further inquiries, email us with the subject line “Inquiry about the iFlytek Activate Program”. 

“Shall I compare thee to a summer’s day? You are cuter and more restrained.”

Shakespeare didn’t write that—maybe because he didn’t have help from an artificial intelligence-powered proofreading tool made by Chinese search engine Sogou.

The company recently rolled out the Sogou AI Writing Assistant, which it says uses “optimized AI algorithms” for its grammar- and spell-checking features. AI also powers its suggested replacements for terms it identifies as “Chinglish,” an English writing or speaking style influenced by the Chinese language and criticized in China for not being “native.”

Sogou said in a statement that the tool is accurate 96.83% of the time. When I saw that, I felt concerned about my editor’s job security.

Its performance reflects how Chinese tech companies treat artificial intelligence as the technological panacea. It is promoted as a “secret ingredient,” over-hyped for its ability to solve problems beyond its capabilities.

I tested the tool using an article I wrote last week about Tiktok CEO Kevin Mayer’s resignation from Bytedance (which had already been edited), and it turned out that Sogou’s AI is so smart that it corrected Bytedance, the company Mayer worked for, to “Bidens.” (Which Bidens? Which Biden?)

The AI algorithms identified the word “Bytedance” as a typo so it suggested an alternative which is a type of daisy, also known as “beggar-ticks.” It also happens to be the surname of a pretty well-known American. 

Let’s try another TechNode story about Chinese electric vehicle maker Xpeng’s US initial public offering (IPO) last week.

This time perhaps the tool didn’t find the term “Xpeng” in its corpus, which the company says contains more than 100,000 English sentences (equal to about 31 books) and upwards of 10,000 pieces of “high-quality human correction data,” so it replaced it with “Peng,” a common Chinese last name. 

What a smartypants.

The tool also changed the phrase “backed by,” which originally expressed that Xpeng had received funds from Alibaba and Xiaomi, to “with the support of,” creating the implication that Alibaba and Xiaomi may have helped it with the IPO.

I then input Shakespeare’s Sonnet 18 into the assistant. 

A Sogou spokesperson declined to comment on my findings but said that the accuracy mentioned in the press release was calculated based on “real-world data from users.”

When the product first launched, local media compared it (in Chinese) to Grammarly—the popular grammar checker software run by a company based in San Francisco, Calif.

To compare, I checked some TechNode sentences with Grammarly. And, to be fair, it also gave me some unreliable suggestions. You still have to review all of its suggestions yourself and make decisions on whether to use Grammarly’s corrections. But overall, it didn’t try to dig into the meaning of sentences, an area that current AI technology is really bad at. 

Here is how Grammarly corrects Sonnet 18.

Grammarly seems to have an issue with the old-fashioned expression of “ow’st,” but other than that it gives Shakespeare a score of 98 out of 100. Well done, William.

According to a Grammarly blog post, it also uses AI for its grammar corrections. Specifically, it uses machine learning, which enables systems to learn and perform tasks by feeding it with plenty of data, and a variety of natural language processing (NLP) approaches, which teach machines to understand and process human language.

Software engineer and tech analyst Ben Dickson wrote in an article published in 2019 that Grammarly’s success was “largely due to its focus on a narrow application of AI NLP: grammar assistance,” because it knows “the limits of current AI technologies.“

According to Dickson, currently, the deep-learning and neural network technologies are terrible at solving “problems that require common sense.”

“When dealing with the meaning of sentences and the semantic relations of words, deep learning algorithms fail miserably,” he wrote.

Accordingly, Sogou’s AI writing assistant fails here. We can see there is a pattern with the tool where it attempts to interpret the author’s meaning and rewrite the sentence using its own logic—which sounds like something that AI could do but is actually just not good at. 

As a result, China has AI teachers, AI doctors, and AI poets—one of which wrote this verse (which disgusted an actual Chinese poet): 

“The rain is blowing through the sea/ A bird in the sky/ A night of light and calm/ Sunlight/ Now in the sky/ Cool heart/ The savage north wind/ When I found a new world.“

I think my editor is safe for now.

Update: This article was edited on Nov. 18 to restore a paragraph removed owing to good, old-fashioned human error. Touché, Sogou AI Writing Assistant.

Revenue for Chinese search giant Baidu contracted 1% in the second quarter compared with the same year-ago period, with the company warning that the downward trend could continue in the second half of 2020.

Why it matters: Baidu narrowed a revenue decline from the first quarter, showing some signs of recovery after the Covid-19 outbreak in China.

• Nevertheless, the company’s ad sales, a major contributor to overall revenue, took a significant hit as companies in China have tightened their belts in the aftermath of Covid-19.

Details: Baidu’s revenue reached RMB 26.03 billion ($3.69 billion) in the quarter ended June 30, down 1% from the RMB 26.33 billion it earned during the same period in 2019, the company announced on Thursday. Baidu previously forecasted sales of between RMB 25 billion and RMB 27.3 billion in Q2.

• Revenue from online marketing was RMB 17.7 billion, falling 8% year on year, as Chinese companies weather the fallout from the Covid-19 outbreak and competition from rivals including Bytedance and Tencent increases.
• Baidu managed to narrow a decline in its revenue in the first quarter, which fell 7% year on year to RMB 22.55 billion.
• Net income reached RMB 3.6 billion, up 48% year on year, slowing from 62% annual growth it saw in Q2 2019.
• The company warned that a decline in overall sales could continue in the second half of the year. During the third quarter of 2020, Baidu expects revenues fall between RMB 26.3 billion and RMB 28.7 billion, representing a 6% year-on-year decline or 2% growth as a result of the “very limited” business visibility in China as a result of the Covid-19 pandemic.

Context: Baidu has plowed billions into diversifying its offerings in search of additional revenue streams. It has focused this spending on artificial intelligence and cloud computing, while the company is exploring enterprise services for growth.

• Nevertheless, a number of these investments, most notably its bet on autonomous driving, have yet to pay off. It’s unlikely the company’s self-driving car project will see significant returns in the next decade, putting additional pressure on its other businesses.

The Chinese government has long pursued a security-centered approach to data. Key regulations such as the Cybersecurity and Encryption Laws place national security at the core of its data legislation objectives. The main goals so far have been to prevent cyber attacks and data leaks by upholding strict technical requirements, and exert control over Internet content and cyber infrastructure. 

China’s draft Data Security Law, issued by the National People’s Congress of China on July 3, marks a shift in China’s data strategy, away from a narrow focus on security and towards a multi-pronged approach balancing security concerns and economic benefits. 

Domestically, it starts an effort to create a regulated market for data as a new commodity. Internationally, it brings data into a coordinated trade strategy, characterized by selective reciprocity and protectionism. 

The draft law means more rules for players in the data space, and will kill off some unregulated trade. But for bigger players, the clarity could make it easier to participate in the market.

A vast, often grey, market 

Tech companies thrive on the collection of data, both in China and abroad. Big data sets can be used to create better drugs, improve manufacturing processes, train face recognition algorithms, and offer targeted content and advertisements, to name a few examples. There’s an enormous market for this kind of data.

China is a particularly big market for data. Tech giants operate thriving data businesses. Small players are experimenting with novel ideas like using data as collateral for bank loan applications. 

According to International Data Corporation estimates, by 2025, the amount of data created, collected, or copied in China will increase from 7.5 zettabytes in 2018 (a zettabyte is equivalent to about 200 billion DVDs) to 48.6 zettabytes, accounting for 27.8% of the world’s data. By comparison, the US will only account for 17.5% of the total data generated globally in the same time period. 

The only problem is that much of this market is at best poorly regulated, and at worst illegal. For employees at large tech companies, hacking servers to acquire datasets is a lucrative side business. This data is then sold in Chinese and international markets. 

Until 2019, data regulation focused on shutting down this illegal trade and regulating what data can be sent out of the country. 

While thriving data businesses emerged, regulators has struggled to govern them. Privacy legislation has evolved significantly since 2017, but they are only aimed at protecting individuals’ personal information. The big, anonymized datasets that are traded to fuel algorithms remain ill-defined and poorly managed. 

A new ‘factor of production’ 

It was only last fall that Chinese economic planners began to publicly recognize the potential of data trading as a legitimate business—and make concrete moves to regulate and develop this industry. 

In November 2019 the Fourth Plenary Session of the Nineteenth Central Committee of the Communist Party characterized data as a “factor of production.” Four months later, experts at the National Information Center compared data to oil:

“If oil is the core resource in the era of industrial economy, then data is the most important strategic resource in the era of digital economy.”

—National Information Center

The Draft Security Law is the first step in an effort to make China the Saudi Arabia of data. 

According to the National Information Center, China will pursue “asymmetric advantages in this new round of global technological competition.” 

Planners aim to take advantage of China’s unique population size to enable its companies to gather, use, and sell unprecedented amounts of data, both domestically and internationally.

Definitions for the new data economy

The draft Data Security Law lays out a vision for a system that replaces a wildcat market with a legitimate and consolidated industry. It promises that data will be traded “orderly and freely” as a commodity in the new digital economy.

It also provides a legal basis for “Internet+” digital government initiatives, calling for the “construction of e-government” and enhanced “capabilities to use data to serve economic and social development.” 

The draft law takes the first step toward this new data economy: It defines what “data” is. It distinguishes between “information” (xinxi) and “data” (shuju), defining data as “any record of information in electronic or non-electronic form.” 

“Information” belongs to the user and is the focus of privacy protections. By contrast, the legal framework governing “data” is multifaceted. It aims not only at protecting individuals’ privacy, but promoting cybersecurity practices and economic development. 

This definition opens the door to legal ownership of big data products, and brings the exchange of such products under a legal framework distinct from that of privacy. But it leaves many technical details about data ownership, pricing of data, and evaluation of data assets to be resolved later. 

Most important among these unresolved questions for ushering the new data economy is standardization.

Since 2014, big data trading platforms in China have experimented with hosting data transactions. Each uses a different technical framework, which makes data exchange between the platforms difficult. 

Similarly, data generated by government agencies across China exists in silos. Recent difficulties in uniting local health code databases during the Covid-19 outbreak illustrated the challenges of centralized data management.

Exactly how this standardization will take place and what aspects of data collection and management it will make homogeneous is left ambiguous. 

Balancing opening up with protectionism

The draft law notes that international data trade is encouraged, but does not offer provisions to make it easier. Rather, it opens the door to restrictions based on economic protectionism.

There is a lot of money to be made by selling data abroad—but China also wants to keep the most valuable data assets in the country. In March, the National Information Center article mentioned above complained: “a lot of high-value scientific data has not been fully shared and used at home but has flowed abroad.”

The 2017 Cybersecurity Law already restricts outbound cross-border flow of “critical data,” defined relative to national security considerations. 

READ MORE: Dust has yet to settle two years after China’s landmark cybersecurity law

With the draft security law, China adds an economic dimension to data protectionism. The new law defines a new category, “important data,” which is assessed not only in relation to national security, but also economic and social development. 

Under this law, data that is deemed valuable for China’s economic and social development would go through cumbersome export controls. A pharma company doing clinical trials in China, for example, will likely have to store all patient data in China and request approval from local authorities before transferring this data to headquarters. 

We don’t know how this would be implemented, but a draft regulation (in Chinese) that has been under review since 2017 is a likely model. It would require companies planning to transfer “important data” overseas to seek approval from local Public Security authorities.

Bargaining chip

The draft also contemplates using data access as a point of leverage in trade disputes. It warns that China will adopt “corresponding measures” (author’s translation) in its trade of data assets. 

Any country that adopts prohibitions or restrictions on Chinese investment or trade on the basis of China’s data practices could be subject to this tit-for-tat data diplomacy. 

If you don’t want to give your data to China, China won’t give its data to you. If you don’t invest in China’s tech industry because you don’t like how China uses your data, maybe you won’t get its data either. 

In practice, China might restrict data export to many Western countries under this clause. The EU, for example, has agreements with “like-minded” countries such as Japan, Canada, and Israel, which stipulate free data flows.  

No such deal has been signed between the EU and China. China is not considered “like-minded” and so the flow of European data into its borders is restricted. Under the new law, the flow of China-made data into Europe would be reciprocally limited. 

The law would also empower the Chinese government to use data flows to retaliate against the United States if, say, it decides to ban TikTok over fears that the app appropriates and misuses consumer data.

Data will increasingly be used as a bargaining chip in trade negotiations.

Decentralized decision-making 

On the flipside, the law opens room for flexibility by putting regional and departmental governments in charge of deciding what constitutes “important data.”

While data export restrictions on national security grounds will likely be controlled at the central government level, other kinds of data will be subject to provincial and industry-specific decisions. 

We will likely see provinces compete to attract companies by lowering restrictions on digital trade—just as some free trade zones currently experiment with tax incentives. The southern province of Hainan has already announced such a program (in Chinese).

Ultimately, whether the draft law will result in a more or less restrictive environment for digital trade in China remains to be seen. Much depends on implementation by central and local governments, which we will see in the coming months and years.  

Currently, the country is at the bottom of the OECD’s Digital Trade Restrictiveness Index, which measures policy restrictions to digital trade in 64 countries in the world. The new draft law is unlikely to change this overall ranking. 

But foreign companies and governments need to be aware that China is setting the rules for its data economy, looking to data as a bargaining chip, and leaving room for relaxed local restrictions. The legal landscape for data transactions is rapidly and drastically changing. 

The head of AI research at Bytedance is leaving the company as the tech giant faces intensifying international scrutiny of its popular short video app Tiktok, Reuters reported.

Why it matters: Bytedance has seen a slew of setbacks in its international operations as countries including the US and India look to limit its presence within their borders.

• The company recently hired former Disney executive Kevin Mayer as Tiktok’s chief executive officer in a bid to assuage concerns over the company’s Chinese ownership.
• China has ambitions to become the world’s forerunner in AI technologies, but it battles to retain its AI talent, according to report by US think tank Macro Polo.

Details: Ma Wei-Ying will leave Bytedance this week, Reuters cited a source as saying. A replacement has not yet been made public.

• The AI researcher joined Bytedance in 2017 after leaving Microsoft Research Asia, becoming its vice president and head of the company’s AI lab.
• Ma was trained in Taiwan and the US, and received a doctorate in electrical and computer engineering from the University of California, Santa Barbara in 1997, according to his Linkedin profile.

Context: Bytedance is facing heightened pressure after the app was banned in India—the company’s largest international market—in late June and faces the possibility of similar measures in the US.

• Tiktok was one of 59 Chinese apps banned in India over national security concerns. Previously, the app was pulled from app stores in the country after a state court found that it was used to spread pornography and encouraged predatory behavior.
• In the US, two federal agencies are investigating whether the app breached a deal it made in 2019 to protect children’s rights.

Experts explained how Singapore is using AI to make the nation smarter at today’s World Artificial Intelligence Conference (WAIC). It was the first time the annual event focused on the city-state, with a special WAIC Singapore stage.

Here’s what happened in the six sessions:

1. Singapore’s national AI strategy

Singapore’s big push into AI is driven in large part by the government, which has made developing and using AI technology a core part of its strategy for the years ahead.

Chng Zhenzhi, director of the National AI Office within Singapore’s Smart Nation and Digital Government bureau, said the goal is for AI to “solve real-world problems.”

She added, “Our goal is to harness the potential of AI to create social and economic value,” which means focusing on what’s needed, what’s practical, what’s ethical, and avoiding technology for its own sake.

There are three pillars to this very practical strategy: digital economy, digital government, and digital society.

Singapore has outlined five projects where AI-related technology will be put to use, Chng explained:

1. Logistics: e.g. intelligent freight planning
2. Municipal services: e.g. a chatbot for reporting local issues
3. Healthcare: applications include personalized risk scores to help with early detection
4. Education: e.g. creating personal syllabi personalized to each child’s strengths
5. Border clearance: e.g. supporting completely automated immigration points

So far, Singapore has committed around $360 million to AI-related research. The next steps will include forming public-private partnerships so that the new, people-oriented AI technologies can be commercialized, Chng said.

2. AI research meets reality

Next up was Li Xiaoli, principal scientist and head of machine intellection at Singapore’s government-funded Agency for Science, Technology and Research (A*STAR). The organization was formed to figure out how exactly all those concepts and real-life needs can be transformed into working tech.

Of A*STAR’s 4,000 engineers and support staff, Li’s team of 120 are dedicated AI and data scientists. They have already set up 10 joint labs in partnership with other government agencies as well as major corporations such as KPMG, DBS, and Singapore Airlines.

Li gave the example of procurement fraud, a major issue for the tax department. With AI, his lab has been able to build what he calls an anomaly detection engine that can assemble a “suspicion ranking” on procurement deals so that auditors can prioritize which ones to check.

It saves time, reduces costs, and increases auditor productivity, he added.

His second example is one that might save lives. In collaboration with Singapore Airlines, the A*STAR scientists have used AI to build a system, using tons of sensors, to predict when aircraft components will have problems.

It shows maintenance crews the probability of failure for specific parts of a plane, avoiding lengthy and expensive downtime for the airline company by preemptively identifying parts that will need to be replaced or repaired.

3. Accelerating AI adoption in Singapore

Continuing with the theme of collaboration and implementation, Laurence Liew, director of AI innovation at AI Singapore said that his organization is what he calls a “tripartite partnership.” That means it’s a government-backed entity—funded by the National Research Foundation—that connects industry with expertise in academia and research institutes.

Liew said AI Singapore does three main things:

– AI research
– AI prizes and challenges
– Catalyzes the application of AI into industries

Because industries need to move fast, the joint AI development projects produce a working product in just nine to 18 months, he said.

Liew gave the example of a public housing group in Singapore that wanted to use AI to predict when its elevators would require maintenance. Since it runs 10,000 buildings with around 25,000 elevators, and many of the buildings are decades old, it was far from a simple project.

AI Singapore got the project rolling by signing up a professor at Singapore Management University (SMU) to develop algorithms that can predict up to a week in advance when an elevator will malfunction.

That’s just one example of AI Singapore’s 20 real-world, AI-driven projects that have been deployed since its inception three years ago.

When the organization works with private enterprises, it contributes up to $180,000, which the company matches. Then the cash is pooled and AI Singapore engages a research institute or university to start developing the tech.

4. Data sharing as a virtue

“AI is fueled by data. Without data, there is no AI,” said Ng See-Kiong, a professor at the National University of Singapore’s (NUS) Institute of Data Science and Department of Computer Science.

Ng posed the question during his presentation: how can governments make their cities smarter when data tends to be trapped in silos?

While there is no clear solution yet, Ng outlined how his research is geared toward getting data to intermingle between the three main data holders: individuals, governments, and companies. The issue is complex because there are often barriers to sharing data within corporations or governments, even between departments.

Right now, Ng is looking into six aspects of this quandary:

1. “Federated learning”: where owners still hold the data but make it available
2. “Machine unlearning”: which is like Europe’s “right to be forgotten” data privacy laws
3. AI ethics
4. Fair data valuation
5. Safety, security, privacy
6. Model-centric sharing

5. AI that is accessible as well as smart

Renee Lo, general manager for data and AI at Microsoft, emphasized AI accessibility in her presentation. The tech giant’s Azure cloud engine is central to this goal, where, in Lo’s words, all developers can access AI services whether they have formal AI training or not.

Examples include Microsoft-developed AI for transcribing audio, detecting or redacting faces, and moderating content for children.

Giving real-world examples of Azure helping developers and cities get smarter, she pointed to Jerusalem, where analysis of public transit ridership led to changes that reduced travel time by 47%. And as commuting became more pleasant, the city’s transit usage went up four-fold.

Another example was in Miami, where sensors monitor the flows in the city’s 10,000-plus kilometers of water pipes to quickly identify leaks and spills. Repair crews are sent out proactively, rather than waiting for leaks to cause visible damage which residents then report.

6. AI in the years ahead

Where do Singapore’s AI experts see the technology headed in the next few years?

The WAIC Singapore stage ended the day with a panel discussion, where a couple of our experts returned—Li and Ng—and were joined by Alibaba Cloud’s Yang Kan. The moderator was Zhu Feida, associate professor of information systems from SMU.

The discussion started with new priorities for AI. Li said that AI learning with less data is the key “fundamental challenge to be tackled.” Fraud detection, he said, is a prime example of the many areas where there’s a strong need for AI-powered tech, yet there is little real-world data from which to build models.

Ng agreed that there is a need to adapt to less data. He added that other forms of AI beyond data and models are needed, such as logical reasoning or creativity, to more closely mimic the function of the human brain.

In the near future, Yang views the improvement of AI tech platforms as crucial to the technology’s accessibility, particularly for platforms geared toward non-AI specialist users, who can then train their own models.

Zhu anticipates that researchers and companies need to better contend with the issue that data is often not free and therefore need to do a better job of pricing and auditing data.

Editor’s note: This article is first published on TechNode Global by Steven Millward, in partnership with WAIC 2020.

Determined to put forward new schemes and applications of AI, the 2020 World Artificial Intelligence Conference focuses on the development of AI technology, especially its industrial transformation from a global perspective.

To emphasize the importance of globalization, the second day of the event is designated International Day. All activities held on this day focused on enhancing global communication by gathering international organizations, government departments, AI enterprises, and innovation projects from all over the world to discuss progress on AI development.

WAIC International Day was officially launched July 10 at the World Expo Center in Shanghai. The opening ceremony was led by Rong Zhiqin, deputy director of Shanghai Municipal Commission of Economy and Information Technology, accompanied by Zhao Xiaolei, the director of UNIDO ITPO Shanghai; Cai Jiong, Partner of ZD Smart Capital and two entrepreneurs: Thomas Knoop, CEO and co-founder of Livecom and Davide Calì, CEO and founder of Expand. The guests came on stage and lit a world map to officially kick off the WAIC International Day. The ceremony was followed by an opening speech by Xu Kunlin, the deputy mayor of Shanghai.

This year, WAIC successfully established on-site forums at three overseas countries: South Korea, Singapore, and Germany, receiving nation-wide and international attention. The South Korean forum has guests from the government and research institutes to share their knowledge on the progress of Chinese and Korean AI enterprises and how the collaboration between the two countries leads to high-quality economic development in the near future.

At the WAIC 2020 Singapore Stage, Chng Zhenzhi, the director of the national AI office of Smart National and Digital Government Office (SNDGO), shared in-depth information on the application and development of AI technology as part of the country’s economic strategy during the Singapore forum. Singapore is one of the world’s most active countries in embracing emerging technologies. Prepared to enter a new era of digitization, it has classified artificial intelligence as one of its four indispensable cutting-edge technologies. As of now, companies and institutions in Singapore have gradually introduced AI into their daily life. With continuous enhancement and computing power, AI can soon be applied to more extensive fields, whether it is in new retail, health care, education, or financial services.

Held at the SAP Data Space in Berlin, the WAIC 2020 Germany Stage is organized by the China Brücke, the German Artificial Intelligence Research Center (DFKI), the German Chinese Chamber of Commerce (CHKD), the Hamburg Port Logistics Joint Stock Company (HHLA), the German-Chinese Artificial Intelligence Association (GCAAI), and SAP. Attracting nearly 300 exceptional participants, the seminar was attended by seven most influential scientists, politicians, and entrepreneurs in the field of AI to conduct in-depth discussions on Sino-German AI cooperation. Since 2013, China and Germany have reached a consensus on the technological partnership, which laid a good foundation for exchanges and cooperation between the two countries. This high-end digital dialogue on artificial intelligence will become an excellent platform to further deepen the partnership, as a new round of global technological revolution and industrial change has arrived.

Leading up to the WAIC International Day, WAIC organized other events to promote international communication. These included the International Smart City Summit, a roadshow of the UNIDO ITPO network, the China Israel AI Industry Investment Summit, and the International Salon of AI innovation and talent exchange. The results were fascinating, as more than 150 AI innovation projects from all over the world participated, including many newly emerged products. Furthermore, 20 high-quality innovation projects from China, the US, the UK, Canada, Turkey, Singapore, Italy, and other countries conducted online and offline exhibitions.

These projects can be applied to extensive fields, including urban management, business finance, agriculture, industrial automation, robotics, data security, and other cutting-edge fields. Apart from the opportunity to face a global audience, UNIDO ITPO Shanghai has also launched a special project competition, where the winning project could be displayed at the 2020 European Open Science Forum in Italy and the 2021 World Expo in Dubai.

With an open mind and an international vision, Shanghai is determined to embrace talent from all fields of artificial intelligence at home and abroad. The 2020 WAIC is the door to global industrial development and social progress as technological innovations from all over the world compete for the spotlight.

One of China’s leading disease control experts has said that that AI could play a larger role in managing future outbreaks in China, particularly in allocating scarce medical resources and identifying early signs of an epidemic.

Why it matters: At the height of the Covid-19 outbreak in China, hospitals around the country faced medical equipment shortages. Medical institutions appealed to the public for donations to fill the gaps left by the government.

• Chinese officials have been criticized for ignoring warning signs in the central Chinese city of Wuhan, despite evidence of community transmission early in the outbreak.
• The country has set goals to become a global frontrunner in AI by 2030, surpassing the US and countries in Europe.

Details: China should use big data to build an AI-based early warning system that can identify threats like Covid-19, Zhang Wenhong, director of the infectious disease department at Shanghai’s Huashan Hospital, said at the opening of the World Artificial Intelligence Conference on Thursday.

• The disease expert said that if AI had been able to identify hallmark features of the virus in December, the medical community may have been able to sound the alarm earlier.
• Nevertheless, Zhang said that data siloing in hospitals will pose a challenge to AI systems.
• In dealing with Covid-19, China imposed measures that effectively locked down more than 60 million people. Instead of AI, Zhang attributed China’s ability to control the virus to “traditional wisdom and solid urban management.”
• Zhang also said that in the future, AI’s role in resource allocation should be expanded.
• “When we sent our medical team to Wuhan we hoped that AI had been deployed across the country to know what we had and what was lacking in Wuhan. This was not the case,” he said.

Context: Zhang became a popular figure during the Covid-19 outbreak in China, leading Shanghai’s offensive against the disease and providing guidance to cities around the country.

• The doctor rose to fame in late January after telling the media that he had assigned doctors who were also Communist Party members to work on the front lines in Shanghai.
• Zhang has since become a well-respected figure in the fight against Covid-19, known for his straight talk and outspoken views on the outbreak.

Hosted by the Shanghai government, the annual World Artificial Intelligence Conference (WAIC) is a global platform for innovative AI startups and corporates to exchange ideas in pursuit of humankind’s greatest technological achievements.

See: World AI Conference 2020 goes live on July 9 – 11

WAIC will host its first-ever Singapore-focused session in partnership with TechNode Global this Friday. The WAIC Singapore stage will be broadcast on July 10, 2020, from 2 p.m.–4 p.m. (GMT +8) on its official website. 

TechNode Global is honored to organize WAIC’s first Singapore-focused session. With an aim to deepen cross-border commercial ties between China and the Southeast Asia market, TechNode Global was set up with the commitment to better serve the Asia tech community and forge a closer bond between the different regions.

The city-state is known for being a trailblazer when it comes to embracing emerging technologies. AI is one of four frontier technologies essential to the country’s push for “digitally readiness,” and Singaporean institutions and companies are already putting AI into daily use. The adoption of AI promises powerful capabilities in fields ranging from retail to healthcare, education, and financial services.

Those who attend will have the opportunity to listen to a series of presentations and panel discussion delivered by industry thought leaders and policy makers. Here’s what you can expect.

Singapore’s national AI strategy

The Singapore stage will kick off with a presentation by Dr. Chng Zhenzhi, director of the National AI Office at the Smart Nation and Digital Government Office. She will discuss Singapore’s plans to harnesses AI to create social and economic value in its Smart Nation journey. Also, the nation’s plans to deploy scalable and impactful AI solutions in key sectors such as transport and logistics, smart cities and estates, healthcare, education, and safety and security.

AI research meets reality

Partnerships between the research community, industry, and government are key to the deployment and commercialization of AI solutions. Dr. Li Xiaoli, head of machine intellection, principal scientist at Institute for Infocomm Research, A*STAR, will talk about how A*STAR’s researchers collaborate with government agencies and industry such as Singapore Airlines and KMPG to deploy AI solutions to address real-world challenges.

Accelerating the adoption of AI in Singapore 

While the benefits of AI are clear, its implementation is not simple and there are several major challenges limiting the adoption. Laurence Liew, director of AI Innovation at AI Singapore, will speak about how AISG tackles challenges such as talent shortages and resource constraints through its 100 Experiments (100E) and AI Apprenticeship Program (AIAP).

Data sharing as a virtue for smart city AI 

With the increased use of AI, data has become increasingly important as well. However, data silos limit the abilities of AI. Data sharing is more important than ever. Hear from Prof. Ng See-Kiong, Professor at National University of Singapore on what more can be done to address data silo problems as he explores the challenges and opportunities of data sharing in a smart city. 

Data and AI for smart cities 

On the topic of the smart city, Renee Lo, GM of data and AI at Microsoft will share the work its researchers are doing in an effort to shape the direction for the next generation of AI following its threefold approach: meaningful innovation, empowering people, and responsible AI.

AI and cross-border data flows 

The Singapore stage will conclude with a panel discussion on the topic of AI and cross-border data flows. Dr. Li Xiaoli from Institute for Infocomm Research, A*STAR, Yang Kan, regional head of data intelligence solutions at Alibaba Cloud, Prof. Ng See-Kiong from National University of Singapore (NUS), and Prof. Zhu Feida, associate professor at Singapore Management University (SMU) will discuss their experiences partnering with industry, academia, and government. They will also offer their perspective on open-sourced AI technology. 

Singapore stage agenda

In addition to the Singapore-focused session, this year’s online WAIC consists of one exhibition platform that will showcase products and services from more than 100 exhibitors, an opening ceremony, two plenary sessions, 10 thematic side stages, and many industrial side stages, all including both live and recorded content. 

Get your tickets here to watch three days’ worth of live and recorded content. For more information, visit the official website of WAIC.

China has spent much of this year trying using surveillance tools to figure out where people are. The outbreak of a major epidemic challenged the government to track people’s movements. Local governments need to know who’s arriving on trains from epidemic hotspots, and, inside cities, who has been exposed to super-spreader events like the one at Beijing’s Xinfadi market.

The country has millions of surveillance cameras, and its companies are world leaders in face recognition. But this technology has been missing in action in China’s contact tracing efforts, while companies like Sensetime and Hikvision have seen their revenues hit during the pandemic.The video cameras that line China’s streets seem to have played no role in contact tracing. Systems to monitor social media turned out to be ineffective. Covid-19 simply ran circles around China’s surveillance apparatus.

Bottom line: China has been building a camera and face recognition-based surveillance system for years and this should have been its time to shine. Instead, star surveillance companies have seen revenue decline as contracts are delayed or canceled. As a new generation of infection tracking systems are rolled out, authorities are relying on a totally different model pioneered by consumer-oriented tech giants Tencent and Alibaba.

Get one client, make it big: International media love to tell stories of China’s facial recognition-driven panopticon. As overblown as they might be at times, they’re kind of true—and they’ve bred a huge market for facial recognition and surveillance equipment companies.

• In 2013, at the annual meeting of the National People’s Congress, the country’s legislature, lawmakers announced that spending on domestic security would eclipse that of the military.
• Local governments came through. In the southwestern city of Chongqing alone, there are nearly 3 million cameras—168 for every 1,000 people—according to figures from research firm Comparitech.
• On the back of government spending, Chinese camera manufacturers have made it big. Out of Hikvision’s RMB 57.66 billion ($8.16 billion) revenue in 2019, about 70% came from China, and 40% of that came from government-related contracts, according to its annual financial results. That document does not make it clear how much of that comes from surveillance equipment.
• Facial recognition algorithms to make sense of all that video data have also raked in some big bucks.
• Sensetime has grown to become one of the most valuable AI companies in the world. The company has received backing from Softback, Qualcomm, Alibaba, and IDG Capital, among others.

Covid-era losses

A bad year: We would have expected surveillance and face recognition to hold up as Covid brought China’s economy to a halt, especially as contact tracing proved key to reopening. But the industry proved anything but immune.

• Hikvision saw its operating income decline by 5% annually to RMB 9.4 billion in the first quarter. It did not provide details to explain the fall in revenue.
• Dahua Technologies saw its revenue plummet during the same period The Chinese surveillance giant’s income fell by nearly 20% to RMB 3.5 billion. Like Hikvision, it did not explain the fall.
• Megvii was reportedly one of around 300 companies that applied for bank loans to soften the blow of the epidemic. In February the company sought RMB 100 million to bankroll technology development, Reuters reported.
• Megvii, Sensetime, and Yitu are not publicly listed, and are not required to disclose financial information.
• More than 70% of security-focused companies in China saw flagging orders in the first quarter, according to a survey by information platform China Public Security (CPS). According to CPS, 7% of security enterprises have had orders canceled

What explains the losses? It’s not clear. A plateau in government contracts and US-led limits on exports probably both contributed. The biggest contributor to losses was likely falling public spending in the first few months of the year. After years of rapid growth, China’s investments in video surveillance have largely remained flat in 2020 in spite of the massive health crisis in the country.

Aside from a single large investment in the southern island province of Hainan, in the first half of 2020, proposed government investments in video surveillance projects around China amounted to RMB 2.7 billion in nearly 160 tenders, up marginally from RMB 2.5 billion in 2019, according to a TechNode analysis of public procurement data.

• Stagnating procurements offer one explanation for surveillance companies’ recent troubles. Some of the country’s biggest AI companies—including Sensetime, Megvii, and Yitu—can largely attribute their success to government surveillance projects.
• Revenues this year are driven by contracts awarded last year. In the short, the question isn’t how much is promised but how much is actually paid. Slowing procurement is likely related to reports of delayed projects.
• Security companies have also seen significant delays to projects for which they have won tenders, according to CPS.
• Delays in large construction projects may have contributed to revenue declines for some companies, as purchases of surveillance equipment were put off along with construction following citywide lockdowns.
• To compound these issues, surveillance companies have also suffered as a result of the geopolitical tug of war between the US and China. The US placed Hikvision, Dahua, Sensetime, Megvii, and Yitu on its so-called Entity List, effectively barring them from doing business with American firms.

Bright spots: Two Chinese provinces stood out in procurement data as exceptions to the trend of falling surveillance contracts: Hebei, the province that surrounds Beijing, and the southern island of Hainan both bucked the trend.

• In 2020, 49, or around a third of all surveillance tenders, focused on Hebei. Guangdong followed, with nearly half that number.
• In 2019, the vast majority of these tenders focused on projects in Guangdong. Beijing and Shandong came in behind the southern province.
• It’s unclear why Hebei saw a peak in surveillance projects.
• One possibility is the province’s close proximity to Beijing. An effective surveillance system could act as a buffer between the capital and the rest of China and may provide more protection from subsequent waves of infections.
• The province has also been working to build a system to enforce anti-straw burning laws, prohibiting open-air incineration of straw waste.
• In Hainan, one tender worth RMB 45 billion for a project focused on provincial video surveillance sharing was made public. It is unclear whether this project has been started.

Missing in action

The dog that didn’t bark: The most surprising thing about the video surveillance industry in 2020 is what it didn’t do. As Covid-19 spread across China, the government went into crisis mode, launching efforts to track and control mobility. Despite tens of millions of cameras deployed across the country, officials turned to pen and paper to track where people were going and where they had been.

• China initially responded to the outbreak by locking down cities—stopping people from moving at all rather than tracking them.
• Travel that was allowed after the lockdowns was governed by paperwork. Train stations handed out paper health declarations as travelers returned home to their loved ones.
• Officials were given access to telecoms data including location information, which proved largely ineffective in contract tracing, only giving estimated locations of a mobile phone’s owner.
• In some cases, this data wasn’t enough to track down people government officials believed to be infected. “We did identify one man from Hubei [on the list from telecom operators]. We searched everywhere for him but just couldn’t track him down,” a cadre from a town in the southern province of Guangdong told the Financial Times.
• TechNode was unable to find any evidence that video surveillance played an important role in tracking transmissions.

Thermal cameras: Surveillance companies in China ended up deploying thermal surveillance systems in public areas. These platforms were touted to be able to measure the temperatures of dozens of people at once.

• These systems didn’t really catch on. They appeared sporadically in malls, schools, and subways, but the millions of checkpoints that government mobility mostly stuck to the infrared thermometer guns used during the outbreak of SARS in the early 2000s.

A new system emerges: As China re-opened, it built a new surveillance system to track where people went. Instead of turning to the camera companies, these contracts went to Alibaba and Tencent.

• Tencent and Alibaba developed “health passports”—mini-apps that assign each user a rating based on their health status and travel history.
• The platforms were completely novel and functioned both as an indicator of an individual’s risk of being infected but also as a way to track movement, as users were typically required to scan QR codes when entering public areas.
• Cities around China were not required to adopt the platforms, but Alibaba said in March that more they were being used in more than 200 cities.

READ MORE: How China is using QR code apps to contain Covid-19

Does video surveillance have a post-Covid future?

Companies like Hikvision and Dahua seem determined not to let the same opportunities pass them by a second time. After the virus started taking hold internationally, the two companies began marketing epidemic surveillance products abroad.

Dahua has reportedly sold thermal imaging cameras to Amazon to measure its employees’ temperatures. Sensetime has also narrowed in on temperature screening, deploying its technology at the entrances to airports and subway stations in China. Meanwhile, Hikvision’s international website is mostly focused on products aimed at adapting to the “new normal,” such as handheld thermal cameras and other thermographic devices

Nevertheless, the failure to effectively contribute to contract tracing raises questions about the industry. If they can’t be deployed to aid in a time of crisis, what are these companies selling, and does it actually work?

Organized by the Shanghai government, the annual World Artificial Intelligence Conference (WAIC)’s mission is to provide a global platform for current and future AI innovation startups and corporates to exchange ideas and resources in pursuit of humankind’s greatest technological achievements.

Unlike the previous two conferences, which were held onsite in Shanghai, WAIC 2020 will be held online from July 9 to 11 due to the Covid-19 pandemic. However, this will not change the conference’s core mission and objectives. This year’s conference will take advantage of its online component by offering a series of online interactive activities and virtual platforms.

With the theme “Intelligent Connectivity, Indivisible Community,” WAIC 2020 aims to gather prominent experts from all over the world to discuss AI technology—its development, its applications, its trends, and how AI can better serve humanity for good.

This year’s online conference consists of one exhibition platform that will showcase products and services from more than 100 exhibitors, an opening ceremony, two plenary sessions, 10 thematic side stages, many industrial side stages, including both live and recorded content. These sessions will be broadcasted from multiple studios in Shanghai, the US, Germany, France, and Singapore. In conjunction, there will be a 3D cloud-based virtual exhibition in an interactive 360-degree virtual environment, and multiple other interactive options for participants to experience AI scenarios online.

10 thematic side stages and featured activities

The ten thematic side stages will focus on three major topics: AI technology trends, the AI economy, and AI homes. Sub-topics include chips, data intelligence, AI + 5G, autonomous driving, AI + industry, AI + blockchain, AI + healthcare, AI + education, and AI governance.

In addition to the thematic side stages, WAIC partners will host industry forums to dive deep into vertical tech topics. Other featured activities include product releases, Global Day, Developers’ Day, the Sail Award, and the AIWIN competition.

During the three-day conference, WAIC has designated the second day to be the Global Day to facilitate global AI project pitching, international talents fair, and a series of roadshows to present global AI innovative products for a network of local and international tech communities.

3D cloud-based virtual platform

The “3D cloud-based virtual platform” is a major innovation in online conferences. Not only does it provide the experience of a traditional exhibition online, it also provides opportunities for digital interaction and activities, increasing touch points between the exhibitors and participants beyond what’s available in an offline exhibition. Participants can choose from a range of interactive methods such as video-conferencing and AI customer service, or join live-streaming sessions with the exhibitors.

The exhibition platform will be divided into seven virtual areas: AI + education, AI + medical, AI + transportation, AI + finance, AI + industry, AI + basic technology, and AI + urban management. In addition to that, a designated AI Pilot Zone will allow key AI corporates to customize a 360-degree virtual environment for their visitors. Adding a little carnival atmosphere to the online conference experience, WAIC has designed several ways for participants to collect AI reward points to redeem gifts, including a treasure hunt.

Speaker lineup

More than 400 prominent guest speakers have confirmed their participation. Among the leaders who have confirmed their participation are:

• Manuel Blum: Winner of the 1995 Turing Award and founder of computational complexity theory
• Dong Mingzhu: The chairman and president of Zhuhai Gree Electric Appliance Co., Ltd.  
• John Hopcroft: Winner of the 1986 Turing Award
• Christian Klein: CEO of SAP
• Li Yanhong: Founder, chairman, and CEO of Baidu
• Li Yong: Director-general of the UN Industrial Development Organization
• Mei Hong: Academician of Chinese Academy of Sciences
• David Patterson: Winner of the 2017 Turing Award and master of Computer Architecture
• Judea Pearl: Winner of the 2011 Turing Award in 2011, father of Bayesian Networks
• Thomas Sargent: Winner of the 2011 Nobel Prize in Economics
• Joseph Sifakis: Winner of the 2007 Turing Award and academician of various countries
• Pascal Soriot: CEO of AstraZeneca Investment (China) Co., Ltd.
• Wang Jian: Academician of Chinese Academy of Sciences and founder of Alibaba Cloud Computing
• Yao Qizhi: Winner of the 2000 Turing Award, Academician of CAS, Academician of American Academy of Sciences
• Zhao Houlin: Secretary-general of International Telecommunication Union

The conference is expecting 150 online exhibitors. At present, in addition to Microsoft, Amazon, Alibaba, Tencent, Huawei, iFLYTEK, ABB, SAIC, and other industry leaders that participated in the exhibition last year, new exhibitors include AstraZeneca, Qualcomm, JD Cloud, Youbixuan, Bank of China and other AI tech companies such as Cerinsi, Cambrian, Yunzhisheng, Suiyuan Technology, and Yunkuang. Stay tuned as the list is being updated on the website.

TechNode Global, the international arm of TechNode, announced its partnership with Chinese AI Giant iFLYTEK on June 8, 2020, aiming to boost the community of thriving startups and AI technology applications across Asia.

Tapping the network of TechNode Global’s tech and innovation ecosystem across China and the Asia-Pacific region, the partnership provides various localized resources of AI technology by iFLYTEK’s open platform for the Asia market. The goal is to foster AI adoption among companies and help them draw on external resources in addition to their regional network.

With expertise and flagship events such as Tech Insights webinar and ORIGIN conferences in the region, TechNode Global will initiate a new AI-focused series of webinars and events in 2020 and will jointly host a Global AI Challenge with iFLYTEK in the Asia region. Meanwhile, iFLYTEK, with its upcoming open platform, will provide numerous resources to the region and dev-ops communities, helping companies in adopting AI into the daily operation of their businesses.

With a vast network in global innovation and entrepreneurship, TechNode Global is a spin-off of TechNode, aiming to be the #1 pan-Asia tech-focused platform with a focus on Pan-Asia cross-border business, events, and tailor-made marketing solutions for startups, VCs, corporates, and other industry pioneers. Being at the center of a unique worldwide tech ecosystem of startups, venture capital firms, industry resources, and corporate partners, TechNode Global supports and connects China’s startup ecosystem with the rest of the world, and facilitates interaction and cross-pollination between China and APAC.

“There is great synergy between iFLYTEK Open Platform and TechNode Global,” said Qi Shuxuan, general manager of iFLYTEK Open Platform. “We hope to strengthen this technology and innovation partnership with TechNode through the real-world application of our AI-powered technologies in this international strategy. The goal is to achieve cross-border integration of AI applications in various fields, generating positive energy and impact. We believe with the combined data, algorithms, expertise, and mutual support from both iFLYTEK and TechNode, we can empower our customers.”

“TechNode Global will continue to play an important role in supporting the tech startup ecosystem in Asia. The global expansion of China’s innovation is beneficial to the overseas innovation environment and TechNode Global platform is committed to facilitate the connection between China and the rest of the world, including Southeast Asia,” said Dr. Gang Lu, founder and CEO of TechNode.

Search giant Baidu has released a toolkit for machine learning that allowing researchers to build and train quantum neural networks.

Why it matters: Tech giants around the globe have increased their focus on quantum computing. Baidu set up its Institute for Quantum Computing in March 2018.

• Meanwhile, Google released its quantum machine learning framework, Tensorflow Quantum, in April this year.
• Microsoft has also pushed to provide quantum computing solutions to the public. In November, the company released a private preview of Azure Quantum, a quantum cloud computing service, with plans for wider rollout in the future.

Details: Baidu’s Paddle Quantum, built on top of the company’s deep learning platform Paddlepaddle, consists of a set of machine learning toolkits, including quantum development tools, a quantum chemistry library, and a set of features for optimization.

• Baidu claims that the toolkit is the first deep learning framework in China to support quantum machine learning.
• The platform allows develops to build simple quantum neural networks or develop their own models from the ground up, Baidu said.
• Paddle Quantum is “more flexible” than similar systems by other companies, according to Baidu. The company claims that it has been able to simplify the implementation of a promising quantum algorithm by 50%.
• Dubbed the Quantum Approximate Optimization Algorithm, researchers believe it could outperform classical algorithms in the next few years.

“From now on, researchers in the quantum field can use Paddle Quantum to develop quantum artificial intelligence, and our deep learning enthusiasts have a shortcut to learning quantum computing.”

—Duan Runyao, director of Baidu’s Quantum Computing Institute

Context: As development of traditional computers reaches its peak, researchers are looking to quantum computing to drive the next wave of artificial intelligence. The technology is based on quantum mechanics and is still in its infancy.

• Nevertheless, Google scientists last year said they have made a major breakthrough in quantum computing. The company claimed that they had developed a quantum processor that could complete in just seconds a computation that would take a traditional computer thousands of years.
• Chinese companies including Alibaba, Tencent, and Huawei have also set up quantum computing research labs.

Chinese internet giant Tencent announced Tuesday it will invest RMB 500 billion (around $69.9 billion) into cloud computing and other technologies over the next five years.

Why it matters: While the Shenzhen-based company said the investment plan is a response to Beijing’s call for investment in so-called “new infrastructure,” it is also strikingly similar to a plan announced by e-commerce giant Alibaba last month.

• Tencent, best known for its instant-messaging app Wechat and mobile games, has in recent years been actively investing in cloud services in search of growth, opening another avenue of competition with Alibaba, China’s biggest cloud-computing service operator.
• The company’s revenue from its fintech and business services segment, which includes cloud computing and consumer loan services, grew 22% year on year to RMB 26.5 billion in the first three months of this year, accounting for nearly one-quarter of its total revenue.
• China is pushing for a technology-driven structural upgrade of its economy with new infrastructure investment projects, which include a wide range of sectors from the next-generation 5G network, artificial intelligence (AI), and data centers.
• Analysts estimated China’s total investment into new infrastructure projects will reach RMB 1 trillion in 2020.

Details: Key sectors of Tencent’s investment plan include cloud computing, AI, blockchain, data centers, supercomputer, and cybersecurity, Dowson Tong, Tencent’s executive vice president, told reporters Tuesday.

• Tencent will build across the country several data centers, each containing more than 1 million servers, Tong said.

Context: Tencent’s announcement came days after Chinese Premier Li Keqiang called for investment in new infrastructure during his government work report speech at the opening of the National People’s Congress on Friday.

• Chinese e-commerce giant Alibaba said last month it would spend RMB 200 billion on cloud computing and data centers.
• Alibaba and Tencent are the two biggest cloud service providers in China. Alibaba reigns supreme with 46.3% share of the market with Tencent holding 18% as of the fourth quarter of 2019, according to market research firm Canalys.

In its latest rebuke against China, the United States said on Friday that it will add nearly three dozen Chinese companies and institutions to a trade blacklist over their alleged involvement in human rights abuses or their ties to the country’s military.

Why it matters: The move comes as lawmakers and political advisors meet in Beijing for the Two Sessions, one of the most important annual events on China’s political calendar.

• Being added to the trade ban effectively bars the companies from buying products and technology from American firms without approval from the government.

Details: The US Department of Commerce added nine companies and institutions to the trade ban for their alleged complicity in human rights abuses in China’s northwestern Uighur Autonomous Region. Of these, seven companies have were added to the so-called “Entity List” for “enabling China’s high-technology surveillance” program.

• These seven companies include facial recognition firms Cloudwalk Technology and Sensenets, surveillance company Netposa, and AI chipmaker Intellifusion, according to a commerce department statement.
• Meanwhile, the US has also blacklisted 24 companies and institutions for their alleged ties to China’s military, the commerce department said in a separate announcement. These include high-profile cybersecurity firm Qihoo 360 and Softbank-backed robotics firm Cloudminds.
• Qihoo responded in a strongly worded statement, saying that the company “firmly opposes” the accusations.
• Several universities and research institutions were also been included on the Entity List for being involved in or posing a significant risk for becoming involved in activities that could affect US national security, including the Harbin Institute of Technology and Beijing Computational Science Research Center.
• The blacklisting aims to prevent China from using US commodities and technologies in activities that undermine the country’s interests, Secretary of Commerce Wilbur Ross said in the statement.

Context: The move marks Washington’s latest offensive against Chinese tech companies in an ongoing spat between China and the US.

• In October, the US placed several Chinese artificial intelligence and surveillance companies on the blacklist, including AI firm Sensetime, facial recognition company Megvii, and surveillance camera maker Hikvision, among others.
• The US has also sought to limit Huawei’s involvement in its 5G rollout, citing national security concerns.

Sensetime, one of China’s highest-profile artificial intelligence startups, is looking to raise $1 billion after seeing wider adoption of the company’s technology during the coronavirus pandemic.

Why it matters: Fresh funding for Sensetime could signal that investors are regaining confidence after the Covid-19 outbreak in China. Money has been scarce for China’s tech startups in the wake of the epidemic.

• Venture capital investments in China dropped by a third during the first quarter compared with the same period a year ago, according to data from business information provider Itjuzi.
• Total investment fell to RMB 119.1 billion ($16.7 billion) in Q1 from RMB 173.6 billion during the first three months of last year.

Details: The $1 billion investment could give Sensetime a valuation of at least $9.5 billion, up from $7.7 billion at the end of 2018, the Wall Street Journal reported, citing people familiar with the matter.

• The round of funding would be the last before an initial public offering (IPO), the people said.
• Earlier this year, Sensetime reportedly delayed its plans to go public, looking instead for up to $1 billion in private funding. The company believed it would see more success in its IPO if it waited, sources said at the time.
• Since the start of the Covid-19 outbreak in China, Sensetime has seen widespread use of its contactless temperature-monitoring technology, which has been deployed in metro stations, office buildings, and factories around China.
• The company declined to comment when reached by TechNode on Monday.

Context: Nevertheless, China’s biggest AI companies have faced increased pressure since several firms were placed on the US Entity list in October.

• US authorities placed Sensetime, face recognition giant Megvii, speech recognition firm iFlytek, and surveillance camera maker Hikvision on the list, effectively barring them from doing business with American companies without permission.
• Sensetime has been moving to make its business more globalized, opening offices or research centers in Japan, Singapore, and the United Arab Emirates.

How should autonomous vehicles (AVs) be programmed to interact with pedestrians? What about non-intelligent vehicles such as other cars and bicycles? While AV technologies themselves—intelligent computing and AI, LIDAR sensors and so on—receive a great deal of government and popular attention, AV technology domestication is an overlooked topic that will influence how fast the industry can develop.

Human behavior changes when confronted with new technologies; we slow down at a speed hump, we smile in front of a camera, and we cross the road when the pedestrian light is green. Science and technology studies scholars call this technology domestication; how do people consume, modify, reconfigure, and resist technologies? Technology domestication is user experience writ large.

As a post-doctoral fellow at the Hong Kong University of Science and Technology, I ethnographically explored this topic in 2019. Over six months, I met and interviewed dozens of people across Beijing, Guangzhou, Hong Kong, Shanghai, and Shenzhen.

My interlocutors worked inside AV and traditional automobile companies as data scientists, engineers, marketers, and strategy advisers, as well as along the supply chain making sensors and other components. I also spoke with industry analysts, journalists, and lawyers.

I did not expect technology domestication to be so top of mind among my interlocutors, but it was. People involved in actually making AVs, as well as those responsible for getting them onto China’s roads, were especially perceptive. In fact, while other countries have been focused on developing AVs that “fit in” with existing behaviors, Chinese researchers are approaching the topic differently.

Yi Zeng, a prominent computer scientist and Director of the Research Center for Artificial Intelligence Ethics and Safety in Beijing, encourages Chinese AI companies to better understand how people will ultimately use and interact with the products and platforms they are creating.

Fitting in

You might think it is the AVs job to fit in with people. A team of social scientists working on AV behavior at Nissan thought just that. They worked hard to develop AVs with “socially acceptable behavior,” defined as behavior that takes into account existing social and cultural practices related to mobility and human-automobile interaction.

Perhaps due to established road rules and ingrained behaviors around Silicon Valley, where the team was placed, Nissan focused on teaching AVs to simulate how a typical driver navigates a vehicle in the presence of others.

Take a pedestrian crossing that does not have traffic lights (i.e., a zebra crossing): even with clear rules of engagement, it is common for drivers to momentarily make eye contact—maybe also using subtle facial gestures—to signal the pedestrian can cross safely. In fact, the pedestrian may let the driver pass first for various reasons; such is the complexity of this seemingly simple human-automobile interaction.

The team proposed that Nissan’s AVs should be equipped with a device that functions analogously to such cultural signaling, alerting pedestrians with a caption that lights up and flashes “I have seen you, you may cross safely.” In this case, the solution ensures that AVs are programmed to behave based on current social and cultural norms.

Standing up to a bully

My interlocutors in China thought differently. After seeing how people treat AVs during testing, they became convinced fitting-in was not enough. Yan Li (a pseudonym), a deep learning (shendu xuexi) engineer at a large Chinese automobile conglomerate developing their own line of AVs, put it pithily when we met in Guangzhou: “Humans bully AVs.”

Also using zebra crossings as an example, Yan Li explained that time and time again during testing, pedestrians crossed the road and payed scant attention to the AV. “They were so confident the AV would stop they completely tuned out.”

This worries Yan Li because outside testing areas in the real world, where traffic is greater and there are more pedestrians, the AV will get stuck. A flashing light alerting the pedestrian they can cross safely is useless. Yan Li explained, “Chinese just won’t let the AV pass. They’ll keep crossing and even loiter, because they know an AV will not harm a human. We need to break the deadlock. But how? That’s our conundrum.”

Yan Li treated the issue matter-of-factly at first; just another algorithm oddity that needed fixing. Over time, however, she came to see the deep social and cultural realities at the heart of the issue. Part of the challenge, she explained, is the sheer variety of users and behaviors on Chinese roads compared to western environments. “What do you do when a chicken crosses the road?” Yan Li asked earnestly. “This is not as uncommon as you think. Chickens move differently to a cat or a dog. We need to think about all these things.”

It’s a fair point; how many chickens roam the roads of Silicon Valley?

Sammy Wang (also a pseudonym), a senior executive at a large Chinese e-commerce company, had similar experiences and concerns. Sammy is part of a team developing autonomous delivery solutions in which an AV travels to the entrance of a residential compound and then dispatches smaller autonomous units to complete delivery to the customer’s door (these smaller units are even capable of riding up and down an elevator). Sammy explained:

During our testing, we had lots of problems. The biggest one was that people would not let the small unit pass them and often ignored it; they just stood there talking or whatever. Even if they noticed it, they didn’t let it pass. It’s a big headache for us.

Engineers were just beginning to recognize this challenge when I spoke to them.

Both Yan Li and Sammy were trying to figure out a way forward: how can AVs operate in such an environment without hurting others? Some people I met believe Yan Li’s and Sammy’s concerns are not relevant, even in China’s unique environment. These interlocutors explained that AVs will anyway be deployed in highly controlled environments as part of China’s smart-city development agenda.

But when AVs eventually share space with pedestrians and non-intelligent vehicles, as is likely, an AV that asserts itself rather than yields may be necessary. But how assertive should it be and what will an assertive AV actually look like?

Will it edge forward and nudge people with a soft yet harmless bumper bar? Will it announce “I am proceeding slowly, please disperse” and move forward? Right now, we just don’t know. It all depends on how people (mis)behave in the future? Yan Li summarizes the challenge nicely: “How can I program an AV to be assertive, yet not endanger others?”

While we’re not sure of the answer to AV technology domestication, at least now we’re asking the right questions.

Tencent-backed cloud artificial intelligence-training platform Enflame Technology has raised RMB 700 million ($98.6 million) in Series B funding, the company said on Thursday.

Why it matters: The coronavirus pandemic has made raising money more difficult for cash-strapped startups. In the first quarter, venture capital investments in China dropped by a third compared with the same period a year ago, according to data from Itjuzi.

• Total investment fell to RMB 119.1 billion in Q1 from RMB 173.6 billion last year.
• Nevertheless, the development of artificial intelligence is a priority for China, with the government setting ambitious goals to become a forerunner in the technology by 2030.

Details: Enflame Technology has closed its RMB 700 million Series B, with Beijing-based private equity firm Summitview Capital leading the round.

• Government-backed Shanghai Innovital Capital, Red Dot Capital, Redpoint Ventures China, Shanghai-based Delta Capital, and social media and gaming giant Tencent, among others, also participated in the round.
• The proceeds of the round will be used to scale Enflame’s business, expand its tech support team, fund research and development, and mass produce its products, the company said.
• Enflame provides products for training cloud-based artificial intelligence platforms, including training and inference chips.
• In December, the company launched an AI accelerator for data centers dubbed Cloudblazer T10. The company said the product provides energy-efficient data processing capabilities for cloud-based AI training platforms.

Context: Enflame is not the only company aiming to improve data center performance. In September, e-commerce giant Alibaba unveiled a self-developed AI chip that was being used to optimize product search and automatic translations on the company’s e-commerce sites.

• Despite being seen as an area in which China can take the lead globally, the country currently lags the US in its development of AI chips and is vulnerable to political actions by US lawmakers, including trade restrictions.

Is China pulling ahead of the US in AI? Not quite, argues Dieter Ernst of CIGI in a recent report entitled “Competing in artificial intelligence chips: China’s challenge amid technology war.” His deep dive into the dynamics behind China’s recent progress on AI chips manufacturing merits closer attention. 

Most western coverage of western AI firms focuses on those that operate in the application layer of the AI stack. But in order for Bytedance to instantly recommend tailored TikTok videos, or for JD to optimize delivery orders, they need to run their applications on hardware.

What AI chips do is optimize performance for specific tasks further down the AI stack. For instance, an AI chip can be tailored specifically for facial recognition, autonomous driving, or cloud computing. The best of these chips represent the bleeding edge of global semiconductor technology, and have grabbed the attention of Washington and Beijing. The Trump administration sees Chinese semiconductor progress as a grave economic and national security threat, and hopes to use a combination of sanctions and incentives to slow down China’s work on AI chips. Beijing officials hope to create a self-sufficient industry capable of withstanding American sanctions and ultimately competing on the world stage.

Read more: China’s AI chip startups—how many will survive?

Basic research lies at the heart of AI development. American researchers invented the field, and have been at its forefront ever since. Ernst contends that America’s “informal, flexible, and undogmatic approach to innovation is, arguably, the root cause for the resilience of the United States’ AI development trajectory.” He argues that “technology diffusion through knowledge networks, combined with intense contests among competing ideas” throughout academia, DARPA projects, and the private sector have made the American AI and semiconductor ecosystem the world’s most vibrant.

In contrast, China has struggled to marry basic research with industry. China’s electrical engineering community was practically wiped out by the Cultural Revolution, forcing researchers in the 80s to start decades behind global best practice. It has to contend with dramatic disconnects between academia and industry as well as, Ernst writes, “the institutional heritage of the Soviet planning system,” which assumed enterprises’ purpose was to meet production targets and not conduct research themselves (that work was reserved for national academies and institutes).

Until recently, the commercial and academic Chinese AI communities rarely interacted. While two of Ernst’s contacts in Chinese consumer-oriented AI companies noted that they had some researchers from public organizations take on moonlight consulting work, these sorts of arrangements pale in comparison to the public-private ecosystem America has created.

Many western analysts have pointed to China’s share of global AI publications as evidence of increasingly successful basic research efforts. The global AI research community is notable for its openness, with academics commonly posting their research in open platforms like Github and arXiv online. “If you don’t share your work, it’s meaningless,” said Yunji Chen, a researcher at the Institute of Computing Technology in Beijing, in a 2019 interview with Nature. 

The global AI community is, by and large, not happy about politics intruding. A Huawei researcher who due to a US State Department travel ban was forced to deliver his presentation remotely received a rousing round of applause at a normally staid conference.

But politics is likely already reshaping the academic community. As Western universities reject Huawei’s money and face increasing scrutiny for connections with the Chinese government’s Thousand Talents plan, western researchers are forced to reconsider their Chinese connections. Co-authorships develop out of connections made through global conferences and academic fellowships, which are less and less likely to be accessible to Chinese nationals. 

I’d be interested to see research that asks whether the global community is splitting into Chinese and western halves—perhaps measured by how often researchers on each side of the divide co-author with the other? This disconnect is likely to harm upstart Chinese researchers more than established western ones.

This comes at a time when basic research is only growing more important in the field. As AI chips are called upon to process massive datasets, companies around the world need to innovate with new architectures. This “paradigm shift” as “the focus of semiconductor innovation shifts from process technology and fabrication to architecture and design at the front end, and post-fabrication packaging at the back end” leaves an opening for a Chinese upstart to contend with American giants like Intel, AMD, and NVIDIA.

But despite the PR bluster, Chinese firms are all to various degrees behind the cutting edge and vulnerable to American actions. Chinese AI chip startups mostly focus on inference, as opposed to training algorithms, a much less technically demanding task. The fact that American capital markets are seemingly closed off to Chinese AI firms is another significant hurdle.  For the time being, going public on Chinese stock markets requires three years of profitability, a rule that discourages investment in R&D.

Industry experts agree that China only really has one player capable of competing with American giants on an even footing in any AI chip vertical: Huawei’s HiSilicon. And even HiSilicon is severely vulnerable to US sanctions. Since its most advanced chips are manufactured in Taiwan at TSMC’s 7nm foundries, American pressure could force Huawei to cut ties with one of its most important AI chip partners. As this week’s recent Department of Commerce regulation release attests, it looks like the US is preparing to drop this hammer.  Given that Huawei has been preparing for years for the US government to come after it, the fact that they still have foreign parts in their flagship phones means that this is much easier said than done.

American firms also stand to lose out from increasing restrictions on their ability to sell to Chinese companies. “A staggering 67% of Qualcomm’s revenue comes from China, for Micron this is 57%, and for Broadcom 49%.” With nearly one fifth of US semiconductor firms’ revenue reinvested into R&D, any big hit to their top line, if not paired with substantial US industrial policy to make up for this gap, will have long term consequences for

US competitiveness relative to Chinese, European, and other Asian competitors.

Chinese companies have made some real progress, particularly in areas where China has a natural comparative advantage. For instance, China currently leads the world in a handful of narrowly defined AI chips related to surveillance. Further, some argue that China has an edge in access to cheap, structured data sources thanks to a large pool of affordable college-educated labor. However, data-labeling is eminently outsourceable, and nowadays Chinese labor really isn’t cheaper than comparably educated Indians or Filipinos. After all, average hourly rates on Mechanical Turk are just $2.

However, American sanctions are forcing Chinese players out of their comfort zone in ways that will help the ecosystem over the long term. As anonymous blogger Youshu writes,

Some will no doubt say that “Yeah, we knew China wanted to develop its own semi industry, so what’s the rumpus?” This observation misses the mark. Before private firms were happy with the Americans, and state firms would just tell their bosses there were no good alternatives. But now orders are being pushed towards domestic rivals, even where they are not very good, providing them with revenues today, and confidence about future revenues, with which to fund R&D.

Ernst is more pessimistic.  He expects “islands of technological excellence [to] continue to coexist with deeply entrenched structural weaknesses in China’s emerging AI chip industry.” The Chinese AI industry will doubtless continue developing, but is unlikely to challenge America for global preeminence any time soon.

A Chinese artificial intelligence firm has become the latest victim of Covid-19 cyberattacks, as hackers look to profit off stolen medical data and research in the midst of the pandemic.

Why it matters: The outbreak has led to a surge in illicit cyber activity, with hackers targeting individuals, companies, and governments.

• Korean and Vietnamese hackers have launched cyberattacks against China and its government in what appears to be an effort to gain intelligence to guide their own Covid-19 responses.
• Targeted attacks through coronavirus-themed emails have increased 667% worldwide, according to US-based security firm Baracuda.

Details: Huiying Medical, a Beijing-based company developing AI-based tools to diagnose Covid-19 infections, had its source code and experimental data stolen, researchers from US-based cybersecurity company Cyble said over the weekend.

• “Credible threat actor THE0TIME” claims to have breached Huiying’s systems and is selling the data on the dark web for 4 BTC (around $30,800), the researchers said.
• The dump includes user data, source code, and Covid-19 experiment information.
• Using images from CT scans, Huiying’s AI system gives healthcare workers a probability that a patient is infected with Covid-19.
• Huiying has cooperated with 800 hospitals around China, including Peking Union Medical College Hospital, Chinese media Leiphone reported.
• Huiying has partnered with smartphone and telecommunications giant Huawei to provide two hospitals in Ecuador with AI-based diagnosis tools for Covid-19, state media reported.

Context: Hackers have capitalized on growing Covid-19 fears around the world as a means to get their hands on sensitive corporate information and personal data.

• Chinese cybersecurity company Qihoo 360 said in February that South Asian attackers used coronavirus-themed emails as bait to launch attacks on organizations “on the frontline” of fighting the outbreak in China.
• Cyble previously found 500,000 Zoom accounts for sale on hacker forums and the dark web. Earlier this month, the researchers also identified a trove of nearly 270 million Facebook accounts up for sale for €500 (around $540).

Chinese facial recognition firm Megvii is reportedly considering a listing on Shanghai’s Nasdaq-like Star Market in order to make the most of favorable policies at home.

Why it matters: Megvii was initially planning a Hong Kong IPO but the company’s inclusion on the US’ so-called Entity List and the coronavirus outbreak have presented significant roadblocks.

• Megvii has been widely expected to become China’s first listed AI startup. The company’s IPO application to the Hong Kong Stock Exchange lapsed in February, even though it had three months to renew it.
• The Shanghai Star Market, launched in June last year, allows unprofitable companies in high tech industries to apply for a listing.

Read more: New STAR Market could lure domestic tech firms from Hong Kong IPO

Details: Despite considering a listing in Shanghai, neither Megvii nor its shareholders are in a rush to go public, one of the sources told SCMP.

• The company is exploring its options and it could still reapply for a Hong Kong IPO, according to the report.
• Meanwhile, Shanghai’s government is seeking to attract high-tech companies by helping them to obtain financing, fast-tracking their IPO applications, and granting them tax incentives—attractive proposition given the difficult first quarter in China.

The Beijing metro system is piloting carriages equipped with cameras that can identify passengers that don’t wear face masks, state-owned Xinhua news agency reported today.

Why it matters: The pilot will contribute to Beijing’s coronavirus response, ensuring commuters follow guidelines about wearing masks in public.

Details: Some carriages running on metro line 6 in Beijing are equipped with high-resolution sensing cameras that transmit video data to an “intelligent system” for analysis, Xinhua said.

• It is unclear what will happen to passengers who don’t wear masks.
• The “intelligent system” monitors drivers to detect if they are fatigued or distracted, prompting them to focus with voice commands if necessary, the news agency said.
• The system also identifies passengers in distress if they are waving for help or fainting, Xinhua said.
• The carriages are equipped with 4K screens showing information about the train’s journey, stops and passenger traffic, Xinhua said.
• The windows turn into screens once the carriage is moving showing “the location, subway map and 3D demonstration of the next station,” the news agency said.
• Stations along the line have been set up with screens showing passenger density and air conditioning intensity of the different carriages on the incoming train, Xinhua said.

Context: Life in China is almost back to normal, after weeks of strict lockdown measures across the country. Authorities and the private sector have been working on technologies to control possible Covid-19 infections whilst normal life goes on.

• These include QR codes that rate individuals’ risk and facial recognition.

Chinese AI startups have raised at least $420 million this month, defying a steep drop in fundraising in China’s tech sector amid the Covid-19 pandemic.

Why it matters: China’s AI firms have grown off the back the government’s drive to become a leader in the technology by 2020 but haven’t been immune to the outbreak of the flu-like virus.

• Nearly all facial recognition startups in China work with the country’s public security bureaus, which can be lucrative given authorities’ emphasis on domestic stability.
• Nevertheless, these contracts have taken a significant hit during the first quarter.

Details: Speech recognition firm AISpeech, AI chipmaker Intellifusion, and machine learning firm 4Paradigm have collectively raised $429 million since the beginning of April.

• 4Paradigm, which specializes in providing machine learning solutions to enterprises, announced on April 2 it had raised $230 million at a valuation of $2 billion.
• Days later, Intellifusion said it had completed a pre-IPO round worth RMB 1 billion ($141.4 million) to fund R & D of its algorithms and chips, among others.
• On Tuesday, speech recognition and analysis startup AISpeech announced its Series C, raising RMB 410 million.
• The investments come after a difficult first quarter. Venture capital funding in China’s tech sector during the first quarter totaled RMB119.1 billion, down more than 30% year on year from RMB 173.6 billion, according to a recent report (in Chinese) by Itjuzi.com. The number of VC deals in the sector fell to 634 from 1,143 during the same period.

Context: Despite signs of a recovery in investor sentiment, the Covid-19 outbreak will likely have a significant impact on China’a AI firms, and the tech sector as a whole.

• In a bid to slow the spread of the virus, Beijing in late-January extended the Spring Festival holiday by more than a week. Even after the extension, many businesses in China allowed their employees to work from home.
• China’s AI companies will likely take a significant battering after the first quarter as their customers, including the government, tighten their belts after a difficult start to the year.

Artificial intelligence company Megvii has open-sourced its self-developed deep learning framework MegEngine, allowing developers around the world to use and improve on the platform.

Why it matters: China has set ambitious goals to be a leader in AI by 2030, but typically relies on US-origin open-source software from tech giants including Google and Facebook.

• The vast majority of Chinese AI companies turn to these frameworks when unable to develop their own. Self-developed systems such as these require massive investments in time and resources.
• Megvii is not the first Chinese company to open-source such a platform for wider use. Search giant Baidu released its self-developed deep learning framework PaddlePaddle in 2016.

Details: Megvii has published an early version of MegEngine, with a beta release expected in June and official launch in September, the company said on Wednesday.

• The framework was initially developed in 2014 as part of Megvii’s Brain++ architecture, which the company uses to train computer vision algorithms—software that allows computers to interpret what they see.
• MegEngine is especially effective when using large volumes of image or video data and can be used for complex tasks including image classification, object detection, and video analytics, according to Megvii.
• Open-source deep-learning frameworks allow anyone with a bit of coding knowledge to train machine-learning models for a variety of purposes without having to build them from the ground up.
• Open-source platforms benefit from people simultaneously using them, working to improve them, and reporting issues for other developers to address.

“We have used [MegEngine] to power our computational photography, facial recognition, and object recognition applications, but the developer community can use our foundational technology in innovative, real-world applications of AI that we have not yet imagined.”

—Sun Jian, chief scientist and head of Megvii Research

Context: China’s AI sector took a hit in October when the US placed eight Chinese companies on the so-called Entity list, effectively banning them from doing business with American firms.

• Megvii was included in the blacklisting. The company later said that its inclusion was due to a “misunderstanding of our company.” AI startup Sensetime, speech recognition firm iFlytek, and surveillance camera maker Hikvision, among others, were also included in the ban.
• Megvii is currently pursuing a listing in Hong Kong, which could make it China’s first AI startup to go public.

Artificial intelligence company Sensetime has delayed its plan to go public in Hong Kong, and is instead seeking up to $1 billion in private funding.

Why it matters: Sensetime has faced numerous hurdles since late last year as it laid plans for its initial public offering (IPO). In October, the US blacklisted the company along with several other Chinese AI firms, effectively blocking it from doing business with American companies.

• The Covid-19 outbreak, which was first reported in the central Chinese city of Wuhan and has now infected more than 200,000 people worldwide, has weakened investor sentiment, causing a massive drop in the number of companies listing in Hong Kong.

Details: Sensetime is looking to raise between $500 million and $1 billion in private funding from new and existing investors or pre-IPO fundraising, the Nikkei Asian Review reported, citing people familiar with the matter.

• The company was working on its listing with investment bank China International Capital Corp. last year but put its plans on hold following the US blacklisting.
• While rival AI firm Megvii previously filed to go public in Hong Kong, Sensetime believes it could see more success in its IPO if it waits, Nikkei sources said.
• “We do not have any new financial plans or information to share at this stage,” a Sensetime spokesperson said in a statement to TechNode on Thursday.
• The company expected its 2019 revenue to increase by 200% to $750 million, Reuters reported in December.
• Sensetime’s valuation reached $7.5 billion in 2019, CEO Xu Li said last year. The company was valued at $4.5 billion in 2018 following its $600 million Series C.

Context: Sensetime was one of eight Chinese companies blacklisted by the US in October over alleged complicity in Beijing’s human rights violations in China. Megvii, surveillance camera maker Hikvision, and speech recognition firm iFlytek were also included in the export ban.

Stonewise, a Beijing-based platform that uses artificial intelligence (AI) for drug development, has raised nearly $10 million in Series A funding, the company said on Monday.

Why it matters: The Covid-19 pandemic has cast a spotlight on the use of AI in treating and diagnosing diseases, and its potential to quickly discover novel treatments for new and endemic infections.

• Covid-19, a new flu-like virus, was first reported in the central Chinese city of Wuhan in late December and has spread around the world, infecting nearly 170,000 people.

Details: Stonewise’s latest round of funding was led by Longhill Capital, with the proceeds being used for research and development, optimizing the company’s technology platform, and expanding its technical and management team, the firm said in a statement.

• Shanghai-based Linear Capital also took part in the round.
• Stonewise was founded in 2018 by Zhou Jielong, former principal architect at Baidu.
• The company has partnered with several domestic and international companies and research institutions including US-based biopharmaceutical firm Beyond Spring, the Chinese Academy of Medical Sciences, and Peking Univerisity’s School of Medicine, among others.
• In February, the company brought on Zhang Yingsheng as vice president of smart drug research and development. Zhang is a former senior scientist at US-based pharmaceutical company Verseon.

Context: With research centers and operations in North America and China, Stonewise is currently looking to expand globally.

• The company is working with the Institute of Materia Medica at the Chinese Academy of Medical Sciences and Peking Union Medical College to research possible treatments for Covid-19.

A little-known Chinese company said that it has developed a facial recognition technology that can identify people in a crowd wearing face masks, as people around the country don the protective gear to reduce their risk of Covid-19 infection.

Why it matters: Tech companies have been working on facial recognition software that can identify people with very little facial data.

• While the technology isn’t new, it has become increasingly relevant as China deals with the fallout from Covid-19, a highly transmissible flu-like virus that was first reported in the central Chinese city of Wuhan in late December.
• Mask wearers have presented problems for facial recognition systems that rely on identifying points around the mouth, nose, and eyes.

Details: Beijing-based Hanwang Technology said that it has developed facial recognition capabilities that can identify every masked person in a crowd of up to 30 people within a second, Reuters reported.

• The company offers two products that use the technology. Its “multi-channel” product uses multiple cameras and can be used for crowds whereas the other identifies one person at a time, at the entrance to a building, for example.
• Hanwang’s facial recognition software is used by China’s Ministry of Public Security, along with around 200 other clients in Beijing, Huang Lei, the company’s president, told Reuters. The company is looking to expand across China.
• Hanwang started working on the technology in January as the outbreak in China began to accelerate, and rolled it out a month later.
• The Covid-19 outbreak has resulted in extra surveillance measures and data collection to track the spread of the virus, a move that has garnered mixed reactions online.

Context: Several Chinese companies have touted their abilities to identify people wearing masks since the beginning of the outbreak.

• Artificial intelligence unicorn Sensetime said in mid-February that it had developed a similar system to Hanwang’s which can identify masked faces in more limited applications, such as identifying employees entering a building. The company uses 240 points on a person’s face to identify an individual, using the parts that are visible to match a face with an identity.
• Megvii in February was reported to be seeking a loan for research and development, which, in part would focus on identifying people wearing face masks.
• Meanwhile, Nanjing-based Minivision was last month commissioned by the city’s government to develop a similar system that can also measure body temperature.

China has released an update to the standards governing personal information, offering new clarity for tech companies including those using biometric data for facial recognition applications.

Why it matters: Companies have been blasted for misuse of data, and even had their apps removed from app stores.

• These standards will apply to personal information collection and usage across the board.
• With more clarity on legal boundaries, tech companies can now proceed with development of facial recognition and other technology.
• “Those who drafted have realized the sensitivity and importance of biometric information, so it receives more protection now,” said Samuel Yang, a data privacy and cybersecurity lawyer and partner at AnJie law firm.

Details: Jointly released by the State Administration of Market Regulation and the State Standards Management Commission, the “Personal Information Security Standards” go into effect Oct. 1, 2020. 

• The latest changes include requiring collectors of biometric data to inform each subject of the purpose, method, and scope of collection and usage, as well as length of time the information will be stored. It also requires that users give express consent.
• The standards recommend storing biometric information separately from personally identifiable information, and refraining from storing biometric information on principle—for instance, deleting original images after extracting the relevant data.
• The previous version said biometric information could be stored if there were adequate technical security measures in place, said Yang.
• There are additional “restrictions on user portraits” and “convergence of personal information collected based on different business purposes” and “management of third-party access.” 
• Companies should not refuse to provide access to functions or reduce service quality if users do not consent, and should obtain their active consent in an itemized way through pop-ups, prompts, or other options.

Context: “These new rules are a reflection of the Chinese authorities’ hands-on enforcement style,” said Yang. “They tend to focus more on how the privacy policy is drafted, if it has necessary clauses, how notice or consent are presented to users,” he added.

• This revision is just one of many regulatory initiatives to delineate what is and isn’t allowed, including laws on personal information protection and data security which lawmakers begin drafting this year. 
• Technological developments like smartphone cameras with resolutions high enough to capture a fingerprint from people making “V” signs raise new questions for biometric data collection.
• Experts have called out apps like Meitu, a popular beautifying app, for excessive collection of biometric data.
• Last year, Beijing Normal University professor Liu Deliang said that much of the legislation on individual biological information consisted of legal vacuums, and lacked measures that could be applied.

Update: added comments from Samuel Yang from AnJie law firm.

China released draft regulations last week, which, if passed, could extend permanent residency to more foreigners, including investors and tech industry workers.

Why it matters: China has a cumbersome employment permit system for foreigners wanting to work in the country, potentially slowing the development of industries it is pushing to get ahead in.

• Officials are concerned about talent shortages in fields like artificial intelligence, which may stymie China’s efforts to get ahead in the tech race.
• China also wants to attract more foreign investment. Making it easier for investors to stay in the country is part of that effort.

Details: The Ministry of Justice on Thursday issued draft “Management Regulations on Permanent Residency of Foreigners,” which includes detailed conditions for prospective applicants.

• Applicants do not necessarily have to be previous residents if they meet other requirements.
• For applicants filing as foreign investors, they must show they have made stable investments for three years, and have good credit history and tax records.
• If they meet that requirement, foreigners that invest more than RMB 10 million (around $1.4 million) individually or through an enterprise in which they are the controlling shareholder, and those that set up high-tech companies in China can apply for permanent residency. 
• Other applicant categories include those that have made internationally recognized outstanding achievements to China’s technology, economy, science, education, culture, or health and sports. In addition, those who fall under urgently needed talent for China’s economic and social development, academic researchers, and management or technical personnel recommended by high-tech companies are also eligible to apply.
• Successful applicants should reside in the country for at least three months of each year. 
• Individuals are able to comment on the draft regulations until Mar. 27.

Context: While some foreigners are already holding permanent resident cards, these regulations clarify and relax some previous requirements for those eligible.

• This is just one move in many the government is making to retool China’s immigration system, and reduce red tape and uncertainty, which could deter tech talent.
• Without international talent, China’s goals of developing tech like semiconductors becomes a lot more difficult.
• “China has an abundance of low- to middle-end labor but is short of talent when it comes to innovation,” (our translation) said Liu Xuezhi, deputy director of the Chinese Academy of Personnel Science. Liu added that it is possible to see, through these regulations, China’s institutional design for attracting talent.

Update: added to context section.

In order to process the CT images of suspected Covid-19 cases in a timely manner, Alibaba DAMO Academy and Alibaba Cloud have jointly developed a new set of AI diagnostic technologies for clinical use. The AI system can accurately analyze the CT images within 20 seconds, with an accuracy rate of 96%. Zhengzhou’s main Covid-19 hospital, launched on Feb 16, has introduced this AI algorithm to aid clinical diagnosis. Zhengzhou is the capital of Henan Province, one of the provinces with the highest cases of Covid-19 outside of Hubei Province.

Nucleic acid testing has been recognized as the main reference standard for the diagnosis of Covid-19 as evidence of pathogenicity. With the accumulation of clinical diagnostic data, the imaging big data features of Covid-19 gradually becoming clearer, the CT-imaging diagnosis results become more and more important. According to the fifth edition of the diagnosis and treatment plan announced by the National Health and Health Commission, the clinical diagnosis does not need to rely on nucleic acid test results. The clinical diagnosis results of the CT images can be used as the criterion for identifying new Covid-19 cases.

The imaging features of the CT chest radiographs of patients with Covid-19 show subtle changes in one or both lungs with patchy or segmental ground glass densities. The number of CT images for one Covid-19 patient is about 300, which puts great pressure on the doctor’s clinical diagnosis. The doctor’s visual analysis of the CT images of one case could take about 5 to 15 minutes.

The medical AI team of Alibaba DAMO Academy based their AI system research on the latest diagnosis and treatment programs, data sampling of more than 5000 cases, as well as Zhong Nanshan and other authoritative teams’ published papers on the clinical characteristics of patients with Covid-19 to come up with a new AI algorithm model to expedite the diagnosis of the virus.

According to reports, through NLP natural language processing of retrospective data and the use of CNN convolutional neural networks to train CT image recognition networks, AI can quickly identify the difference between Covid-19 images and ordinary viral pneumonia images, with a final recognition accuracy rate of 96%. AI takes less than 20 seconds to identify each case on average, which can effectively reduce the pressure on doctors. In addition, AI can also directly calculate the proportion of the lesion site on a patient’s lungs, thereby quantifying the severity of the cases to greatly improve the efficiency of clinical diagnosis.

Editor’s note: This is part of our ongoing Tech for Good series, highlighting how Chinese tech companies are helping fight the impact of the coronavirus. This was originally written by Steven Lee, a writer for our sister site, TechNode Chinese. Read the Chinese version here. 

Artificial intelligence firm Megvii’s application to list on the Hong Kong stock exchange has lapsed six months after the company filed its paperwork for a listing on the city’s bourse.

Why it matters: The Alibaba-backed startup has faced several hurdles as the company attempts to go public. Megvii was blacklisted in the US along with several of its Chinese counterparts, effectively banning it from doing business with American companies without permission.

• Megvii initially filed its application for its initial public offering (IPO) during the second half of 2019 but reportedly decided to push back its listing to this year following additional scrutiny from the Hong Kong board.
• Megvii has also seen delays as a result of the China-US trade war and protests that broke out in Hong Kong last year.
• The novel coronavirus has added additional complications to the company’s plans, as many Chinese businesses came to a standstill in the weeks following the Lunar New Year holiday.

Details: Megvii’s IPO status was listed as “lapsed” on the Hong Kong stock exchange’s website on Tuesday. IPO applications are required to be finalized within six months of their submission date.

• The company plans to refile for a listing in Hong Kong as soon as it has updated its financial data from 2019, two sources close to the matter told Reuters.
• Megvii was not immediately available for comment when contacted by TechNode on Tuesday morning.
• According to Hong Kong rules, companies can renew lapsed applications within three months of expiration.
• Megvii had not disclosed how much it seeks to raise in documents submitted to the Hong Kong stock exchange, but estimates range from $500 million to $1 billion, according to people cited by Bloomberg and Reuters.
• The company in May raised $750 million in a funding round led by Bank of China’s equity arm. The investment valued Megvii at more than $4 billion.

Context: Megvii is one of several companies spearheading China’s AI boom. Others include Sensetime, Yitu, and Cloudwalk, though none have yet gone public.

• Megvii could become the first to tap the capital markets and act as a litmus test for other companies in the same industry to follow.

The US has approved an application by speech recognition firm iFlytek to exempt the company from a months-long trade ban in order to buy medical supplies.

Why it matters: An epidemic of a new flu-like virus dubbed Covid-19 has killed nearly 2,600 people in China after appearing in the central Chinese city of Wuhan late last year. The outbreak has resulted in medical supply shortages around the country with hospitals in several major cities appealing to the public for donations.

• The US placed several Chinese technology companies including iFlytek on the so-called Entity List in October, effectively blocking them from doing business with American companies without permission.

Details: iFlytek applied for an exemption on Feb. 7 to make “charitable donations” of medical supplies that it was restricted from buying in the US as a result of the ban. The US Department of Commerce has subsequently approved the request, iFlytek said in a filing to the Shenzhen Stock Exchange on Monday.

• The company said previously that it had commissioned a US-based law firm to submit a formal exemption request, making an appeal “in the spirit of humanitarianism and international cooperation.”
• The Covid-19 outbreak has put huge pressure on China’s healthcare system. Hospitals in Beijing, Guangzhou, Shanghai, and Shenzhen, among others, have asked the public to donate respirator masks and other medical supplies for healthcare workers.
• iFlytek said its artificial intelligence technology is being used to screen people for Covid-19. The company’s tech is also being used in online learning platforms for house-bound students around the country.

Context: China’s tech sector mobilized its resources to curb the spread of the infection. Meituan, Alibaba, and Tencent, among other companies, have made donations exceeding RMB 3 billion ($429 million).

• Meanwhile, companies including AI firm SenseTime have started offering online classes while students learn at home amid school closures because of the virus. E-commerce platforms have also stepped in. JD.com vowed to ensure adequate supply of face masks to the public and curb consumer stockpiling.
• The US blacklisted iFlytek and several prominent Chinese AI companies including Sensetime and Megvii over their alleged complicity in Beijing’s human rights violations in the Xinjiang Uighur Autonomous Region.

Artificial intelligence firm Laiye has raised $42 million in Series C funding, the company announced on Monday. The round comes as Chinese startups scramble for cash amid a new flu-like epidemic that has swept the country.

Why it matters: Fundraising by Chinese startups has this year slumped by around 60% year on year to $1.8 billion. The number of deals has fallen to 168 from 440 at the same time, according to data from consultancy Preqin.

• The virus, known officially as Covid-2019, has created uncertainty for startups as investors worry about the short-term effects of the outbreak on China’s economy.
• In June, Laiye raised $35 million in a Series B led by cross-border venture capital firm Cathay Innovation.

Details: Laiye’s Series C is co-led by Lightspeed Venture Partners and Lightspeed China Partners, the company said on Monday.

Context: Laiye last year merged with Awesome Technology to launch UiBot, an RPA platform that allows companies to automate rule-based operations that the company claims will “significantly improve operational efficiency,” while decreasing labor costs.

Sensetime has launched a series of free online educational tools allowing Chinese students to learn about artificial intelligence, as schools around the country resort to using online classes amid a virus outbreak that has rocked the country.

Why it matters: An epidemic of a flu-like virus dubbed Covid-19 has killed more than 2,000 people in China. The outbreak has resulted in extended closures for schools well beyond the Spring Festival holiday.

• China’s education ministry on Monday launched a national cloud learning platform that is aimed at providing learning resources for high school students around the country.
• Meanwhile, schools have started hosting classes online and broadcasting them on TV so students don’t fall behind on their studies.

Details: Sensetime will offer complimentary videos for AI-focused online classes, an interactive platform for learning to program and practicing AI theories, and courses for educators to learn how to teach the content, the company said in a statement on Monday.

• The video classes focus on the fundamentals of AI, machine learning, and robotics. The content is available on several online learning platforms in China, including provincial educational cloud platforms and video-sharing site Bilibili.
• Meanwhile, teachers can use Sensetime’s instructor training materials, which include details on “AI development to applications and algorithms.”
• Sensetime is also offering free live-streamed lessons that allow for real-time conversations, the company said.
• During the first semester of the 2019-2020 school year which began in September, 140,000 students from cities including Shanghai, eastern China’s Qingdao, as well as Hong Kong and Macau used the company’s AI curriculum, Sensetime said.

“AI applications have made contributions to the prevention and control of the epidemic—in terms of screening, diagnosing and monitoring the disease through data analytics. With the rapid adoption of AI technologies in various industries, we see a rising demand for AI talents.”

—Lynn Dai, general manager of Sensetime’s education business

Context: Sensetime is not the only company offering free online classes as a result of the outbreak. Vipkid pledged to offer free classes to children between the ages of four and 12.

• Meanwhile, schools around the country have been using Alibaba’s enterprise communication platform Dingtalk and Tencent’s Wechat Work to conduct remote video classes so students don’t fall behind.
• Around 50 million students and 600,000 teachers used Dingtalk to conduct remote classes on Feb. 10, the first day back in class for many students after the extended holiday.

Search giant Baidu has released an open-source tool to detect whether individuals in crowds are wearing face masks, as cities around the country impose rules requiring use of such protection in public spaces.

Why it matters: Authorities in China have taken drastic measures to curb the spread of Covid-19, a new flu-like virus that first appeared in the central Chinese city of Wuhan late last year.

• In Hubei, the province at the center of the outbreak, whole cities have been cut off from the outside as China attempts to contain the virus.
• Several cities including Guangzhou and Beijing are enforcing the use of masks in public areas, including restaurants, shopping malls, and public transport.
• Mask inspections have largely been manual, making identifying non-mask wearers in crowds difficult.

Details: The face-scanning model uses artificial intelligence to identify people in real-time who are not wearing masks or those who are wearing them incorrectly, Baidu said on Thursday.

• The system can identify non-mask wearers with 96.5% accuracy, which meets the needs of routine inspections, according to the company. Developers then only need a small amount of data to train the tool for their own use.
• The model was trained on a dataset of 100,000 faces, Baidu said.
• The system, which Baidu claimed is the first of its kind, can help businesses check if their employees are wearing masks, and also help authorities speed up mask checks in public places.

Context: Face masks have become a necessity in China, where nearly 1,400 people have died as a result of the infection.

• Experts around the world say that hand-washing is more important than wearing masks. Nevertheless, masks are seen as mandatory in China.
• People around the country have opted to isolate themselves in their homes and businesses have implemented work-from-home policies to counter the spread of the disease.
• Baidu recently released another tool allowing people to determine risk levels for Covid-19 infection using the Chinese government’s diagnosis and treatment plan for the virus as well as records from millions of online medical consultations.

Based on the requirement of the detection of fever symptoms caused by the coronavirus pneumonia epidemic, the R&D team at Megvii has introduced an AI temperature measurement solution and rolled it out in the lobby of the Beijing Haidian government building and some subway stations.

This system uses a front-end infrared camera to identify high-temperature people in the crowd. Based on its accuracy of the human body and facial recognition, it can help screen various public places (train stations, bus stations, subway stations, airports) and other high-density areas for abnormal body temperature. This system can help solve the issue of efficiency and controllability in open spaces by using AI-assisted non-contact temperature-sensing measurements during and after the epidemic.

Moreover, under circumstances where one is wearing a mask and a hat, the system is still able to identify the individual efficiently, with a recognition error of ± 0.3 ° C. As such, the public does not need to remove their masks nor hat, or line up for checking.

Another major feature of the system is its long-range temperature measurement capability of more than 3 meters. Once a suspected individual with fever is detected, the system will automatically alert the authorities. Combining this feature with the self-developed human ReID (Re-Identification) detection and retrieval technology, the system can quickly help screen the location clues of fever personnel for further confirmation and medical observation.

A Megvii R&D representative said that the bandwidth of this smart system can screen up to 15 people per second, and can be deployed at 16 exits simultaneously, which enables the screening of all exits at one subway station. In addition to that, with the help of this AI system, only one staff member is required to monitor it, even in the most crowded subway station, which greatly reduces the exposure risk of front-line staff members.

Editor’s note: This is part of our ongoing Tech for Good series, highlighting how Chinese tech companies are helping fight the impact of the coronavirus. This was originally written by Steven Lee, a writer for our sister site, TechNode Chinese. Read the Chinese version here. 

Speech recognition firm iFlytek has applied for an exemption to a US trade ban in order to buy medical supplies amid a national campaign to curb the spread of the novel coronavirus.

Why it matters: More than 900 people have died as a result of the infection, which was first reported in Wuhan, the capital of China’s central Hubei province, in late December. Medical supply shortages have been reported across the country, with hospitals in several cities requesting donations from the public.

• The government has locked down 15 cities in Hubei, effectively quarantining more than 50 million people. Similar measures are now being taken in Zhejiang province, on China’s eastern seaboard.
• The US placed iFlytek along with several other Chinese technology companies and government agencies on the so-called US Entity List in October, effectively banning them from doing business with American companies.

Chinese tech firms brace for impact from coronavirus

Details: iFlytek seeks to make “charitable donations” of medical supplies that it is currently restricted from buying in the US as a result of the ban, the company said in a filing to the Shenzhen Stock Exchange on Monday.

• iFlytek has commissioned a US-based law firm to submit a formal request to the US Department of Commerce “in the spirit of humanitarianism and international cooperation,” the company said.
• The outbreak of the coronavirus has accelerated since mid-January, putting immense pressure on the country’s healthcare system. Major hospitals in cities including Beijing, Guangzhou, Shanghai, and Shenzhen have appealed to the public for donations of respirator masks and other supplies for healthcare workers.
• iFlytek has switched to a non-US supply chain following the ban, the company said, instead using domestic suppliers.

Context: China has mobilized its tech sector in an effort to curb the spread of the infection. Dozens of companies including Meituan, Alibaba, and Tencent have made donations in excess of RMB 3 billion ($429 million), while also deploying their technologies in applications ranging from diagnosis to tracking the spread of the disease.

• Meanwhile, e-commerce platform JD.com vowed to ensure adequate supply of face masks to the public and to place curbs on consumer stockpiling.
• The US blacklisted iFlytek and several prominent Chinese artificial intelligence companies including Sensetime and Megvii over their alleged complicity in Beijing’s human rights violations in the Xinjiang Uighur Autonomous Region.

When it comes to AI and robotics, there are the optimists, and then there are the skeptics. Working in the AI and autonomous transportation vertical, I have struggled to convince people to adopt unmanned solutions. “This is so far away,” they often say. “What is the use case of autonomous drones anyways?” they would ask.

As the outbreak of novel coronavirus becomes an international crisis, the Chinese technology industry is getting creative in finding solutions to avoid human-to-human interactions. Suddenly, these previously “useless” and “too far ahead” technology have found their use cases.

If you can’t see the YouTube player above, try watching here instead.

Unmanned aerial vehicles

Last week, a video of a policewoman using a drone-carried speaker to warn pedestrians to wear masks went viral on Chinese social media. The combination of policewoman + dialect + unmanned aerial vehicle provided comic relief for the nation during the depressing outbreak days.

Not only are drones being used to patrol and promote healthy behavior, but consumer drone company DJI also allocated RMB 10 million (about $1.4 million) to combat coronavirus by donating medical equipment, funding drone-enabled disinfection, and establishing drone-enabled disinfection protocols. DJI competitor and China’s leading agriculture drone tech company, XAG, allocated RMB 50 million to allow drones to be used for disinfection in remote areas.

Agriculture drones have been widely used to spray fertilizers and pesticides since 2013. These well-established drones seamlessly transferred to medical missions to support China’s effort to contain the outbreak.

VIDEO: How tech is changing agriculture in China

Unmanned ground vehicles

Both the Guangdong People’s Hospital and the Hangzhou First People’s Hospital have deployed unmanned ground vehicles to deliver medication and food to quarantined patients. These minimize interaction between nurses and patients.

Each trip, an unmanned robot can deliver four meals with the ability to use elevators, avoid obstacles and find their way back to chargers.

Though these unmanned delivery robots are still in the pilot stage, it has revealed a pain point in the medical field that can be solved by unmanned systems.

Remote work

I am a digital nomad and an advocate of remote working.

A month ago, Zoom stock was falling despite good quarterly results. But the sudden spike in remote work caused by the outbreak has revived Zoom’s stock price. Since corporates in Beijing went back to work remotely on Feb. 3, after the prolonged Chinese New Year, Zoom’s stock price has been climbing. On the day of Feb. 3, Zoom had a closing price of $87.66, with the highest daily percent change of +15% over eight months.

At a hospital built over the past two weeks in Wuhan, Huawei, in cooperation with China Telecommunications Corporation, provided a remote video diagnostic center supported by optical cables. In the future, the company says, the remote diagnostic center will be supported by 5G.

The special circumstances posed by the outbreak have pushed people to use technology in ways that we could not have imagined were necessary a few weeks ago. Technology continues to play a part in the fight against the disease. Not all these technologies will pan out, but they’re getting a real-world test and some will probably emerge with proven applications

This week, TechNode’s translation column gets into the weeds with export controls, with an abridged translation of a deep-dive on AI export controls. TechNode has not independently verified the claims made below. This article was co-translated by Jordan Schneider.

The trade war may (nominally) be at a pause, but the US-China “tech war” is far from over. Over the past several years, both countries have sought to “decouple” from each other, unwinding the complex technology supply chains that bind them together and make them interdependent. The latest salvo from the US was a Jan 6 announcement restricting the export of AI software that can automate geospatial imagery analysis.

This new regulation might sound awfully specific—and it is—but it matters for two reasons. Firstly, it comes on the heels of much buzzed about ongoing review of how the US restricts the export of “emerging technologies.” Despite fears of a sweeping ban on such technologies, this initial restriction has proven remarkably targeted.

As narrow as this ban may have been, Chinese companies are not out of the woods, and it remains unclear what we can expect from the US Department of Commerce. How China’s remote sensing industry reacts to the current export controls gives us a peek into how future export controls may unfold.

Below, an anonymous Chinese analyst argues that while certain industries like self-driving cars may suffer from the export restrictions, the nature of AI research is such that it will not be severely affected unless the US adds considerably more aggressive restrictions.

Swords into plowshares: After the US’s AI software export ban, will an “earthquake” shock China’s remote sensing industry?

By “Tibetan Fox” for Nao Ji Ti, Jan 7.

From the ZTE incident of 2018, to the Huawei saga of 2019, and now to the just-announced AI software export controls of 2020, the increasingly geopolitical international situation has kept our gaze continuously trained on “technologically sensitive” fields.

This time, it is “AI” and “remote sensing” that have been bound together under a ban.

Those friends who keep an eye on current affairs might already know that the new US export controls cover several categories of geospatial imaging software companies, with the fields of UAVs and self-driving vehicles feeling the greatest impact. [Translator’s note: in both English and Mandarin, a variety of terms are used to describe unmanned and autonomous systems, depending on the mode of transport and degree of automation; this article generally translates wurenji as “UAV” or “drone” and zidong jiashi as “self-driving vehicles.”]

In the words of James A. Lewis, a technology expert with a US-based think tank, the Center for Strategic and International Studies, this is “to keep American companies from helping the Chinese make better AI products that can help their military.”

But wait. How is it that artificial intelligence and spatial remote sensing technology are relevant to military matters?

Not long ago, the US demonstrated for the world a “new game mode” for AI-powered remote sensing—the precise “decapitation” of Iranian regional top commander Qasem Soleimani, with his executioner not even being a person, but a Reaper UAV. From a thousand miles away, the drone’s operator lobbed four Hellfire missiles down at him, and boom…

Of course, in China—and an overwhelming majority of countries—AI and remote sensing are more connected with applications such as agriculture, surveying, prospecting, mapping software, and so on. In that case, will these software export controls bring about a “semiconductor-style” crisis [i.e., a “ZTE moment”] for these emerging fields?

Swords into plowshares: the epochal link between AI and remote sensing

Viewed as a whole, the application of AI to the field of remote sensing builds on three core functions:

1. High-intensity, real-time data processing that can handle diverse data sources and structures
2. Effective obstacle avoidance and automated operations that can respond to complex weather and environmental conditions
3. Improving environmental monitoring, decision-making, and early warning capabilities in areas that humans cannot reach

Artificial intelligence adds value in one more way to the domain of remote sensing: namely, by combining UAVs, self-driving cars, and other sensing devices, one can endow endpoint devices with an intelligent “brain,” substituting for humans to complete tasks that would previously have been unachievable.

The AI ban: will an ‘earthquake’ shock Chinese remote sensing industries?

Just how much of a “shock effect” will US export controls have on Chinese remote sensing activities?

Looking at where things stand, it seems as if everybody has finished watching the show and snacking on melons, each wandered home, and got back to whatever each was supposed to be doing. Since nobody has had to whip out a “spare tire” following the imposition of export controls, unlike with the semiconductor industry, no one is now shouting themselves hoarse with criticisms or questions.

Is the role of AI software in remote sensing not big enough?

First, the export controls are on software that automates the analysis of geospatial imagery and will decidedly not cause a quick chain reaction.

The main function of this type of software is in training deep convolutional neural networks to automatically analyze geospatial imagery and point clouds. For example, discerning between vehicles, houses, and other such objects, or reducing pixel variation during scaling, resizing, and other operations.

Overall, when it comes to these AI software restrictions, there just isn’t enough for everyone to start panicking.

On the one hand, the export controls will trigger a relatively long-lasting tug-of-war, and will also implicate a relatively large number of interest groups. These export controls will impact several industries that use the related software as a base to develop aerial photography, 3D maps, and so on, because many of the software products in question are built on platforms such as AWS and GCP that directly provide map processing APIs.

But products built on open-source software like TensorFlow and PyTorch will in fact not be impacted, and affected companies, platforms, and communities can take active measures to avoid this problem.

The most direct example comes from October 2018, when the US Department of Commerce added eight Chinese companies, including Hikvision, Dahua, iFlytek, Megvii, SenseTime, Yitu, and others to its Entity List, stipulating that NVIDIA, Intel, and other companies were not to sell microprocessors to them. But to this day, China remains an important and indispensable market for these US enterprises.

This is not the first time that the US has restricted the export of certain technologies, and it will certainly not be the last. Everyone has become used to barrier after barrier, and their own industries have begun to grow… All one can say is, there is real truth to the name “Chuan Jianguo”. [“Chuan Jianguo” is a nickname for Trump among Chinese Internet users, translating roughly as “Trump Builds the Country.” It is a sarcastic compliment suggesting he is helping China, particularly by engaging in a trade war that forced China to reduce US dependency.]

Of course, the more important thing is that this software “blockade” could, from an objective point of view, create a lag in the development of China’s UAV and self-driving car industries. On the one hand, there are numerous prerequisites to combining spatial remote sensing software and artificial intelligence, such as the practicalities of integrating high-resolution remote sensing image wavebands, specialized microprocessors and datasets for vertical fields, and so on, all of which will impact the accuracy and usability of AI in the domain of remote sensing.

At the same time, China itself has sufficiently strong prior experience and R&D superiority in the field of AI algorithms, with examples such as DJI, the UAV company with the best techniques; Huawei, which has invested much energy into AI edge computing; Baidu, with its R&D capabilities in self-driving vehicles; SenseTime, which has pushed out an artificial intelligence analytic platform for remote sensing; and so on. With all this, export controls should make one less tense.

The most crucial aspect is that the US’s AI software export controls could well “kill 800 invaders at the cost of 1,000 troops.” As everyone knows, the rapid development of artificial intelligence is inseparable from an open atmosphere and industrial environment, and software development in particular relies on open-source, trust, and a global communication environment.

Much of US industry and academia also need China’s strengths to develop projects together, which is why when the US requested that GitHub stop Chinese registrations, GitHub considered opening up a subsidiary in China. [In fact, although GitHub has restricted the accounts of developers from countries under US trade sanctions including Iran, Syria, and Crimea, Chinese users are not affected, nor does it appear that registrations from the affected countries have been halted. However, GitHub appears to still be actively considering opening a subsidiary in China over related concerns.]

At the end of the day, China is an enormous market for AI R&D, applications, and supply chain manufacturing; being “invisible” to China also means China will be “invisible” to onesself, which will only accelerate the birth of Chinese versions of Android, iOS, and Github.

This is also why some web users have suggested that this round of export bans on geospatial software is similar to the 2000s’ “strong cryptography export ban,” in that it is doomed to fail: You can stop a supplier with unique technologies from providing restricted hardware components, but you cannot halt knowledge transmission in an entire field!

Indeed, this time the “knowledge” is already in China, and does not even need to be “exported!”

AI remote sensing: entering ‘No Man’s Land’ as the next step

Of course, this does not imply that China’s intelligent remote sensing technology industry can rest completely at peace.

Even though a particular round of software export controls may not cause serious injury to the entire industry, when viewed as a whole, the field of remote sensing still has many more areas that are not mature.

For example, in the domain of China’s remote sensing, AI inference algorithms have already made considerable strides, but specialized inference chips are still being behind, with most algorithms still employing general-purpose microprocessors produced by NVIDIA and others. As a result, certain gaps in efficiency may emerge when using domain-specific architectures (DSA). Also, the precision level of manufacturing needs to be improved, which affects R&D of top-notch remote sensing equipment similar to the Reaper UAV.

Additionally, having previously raised the importance of data to intelligent remote sensing, it is worth noting that much geospatial remote sensing data is provided by coordination with satellite networks. This creates a particular need for aviation information industries to build systems that can guarantee high spatial resolution, high temporal resolution, high spectrum resolution, and high data quality; for example, the formation of the recent Beidou satellite network has valuable and long-term significance for intelligent remote sensing. And higher-quality, higher-complexity remote sensing data will also require greater computing power to process, further requiring that national semiconductor industries continue to forge on.

Furthermore, remote sensing technology also relies on comprehensive technological upgrades in sensors, monitoring equipment, UAVs, and other fields. For example, the MQ-9 Reaper is “nearly soundless” during flight, and only because of this can it attack its target without being found out before the fact. Looking at the current situation, no matter whether it is the application of satellite remote sensing imagery or the technological threshold of artificial intelligence, all lack adequate relevant professional support, and there remains a need to improve the iterative speed of the integrated process from image collection to understanding to analysis to training and beyond.

When all is said and done, as remote sensing gets more and more developed, and lone warfighting units become more and more integrated with AI, in time we will still see many more creations that bring prosperity to society and humankind. Against this trend, what is needed is not the blind optimism of motivational catchphrases, only that we band together in strength as we face great hardship in breaking new ground.

The small city of Trikala, Greece offers some quintessential provincial scenes: bustling farmers’ markets with vibrant colors and old men with bushy mustaches chatting on park benches.

Delve deeper and you’ll discover public wifi, smart parking facilities, and coming soon, driverless buses. Trikala has become Greece’s first smart city thanks to the roll-out of multiple digital initiatives. With technology delivered straight from China, the city is set to commission (in Greek) the world’s first operational pilot for autonomous buses in real traffic conditions downtown.

Chinese state-owned vehicle manufacturer Weichai will provide the driverless buses which will operate for at least two years. This is the first time that China-made driverless vehicles will hit the roads in Europe.

The buses will automatically avoid obstacles and pedestrians and offer an on-demand service. They will provide customized options for passengers that deviate slightly from original routes to better serve their needs. 5G networks will support operations with lower latency and quicker connection speeds to the control center.

“There was great interest from European manufacturers. Weichai participated through a local subsidiary called Amani Swiss,” Odysseas Raptis, chief executive at e-Trikala, the company responsible for procurement, told TechNode. The most important factor was the technology and know-how of candidates, he said.

The project received funding from the Greek government and the European Union. The two governmental authorities handed out rounds of funding last summer and announced a procurement tender.

A team of five to seven engineers and experts from Weichai will accompany the driverless buses to the city for about nine months. During the first phase, the team will work with local engineers to map out a route. This phase is expected to last two to three months, Raptis said.

The driverless buses will then operate for six months while the Weichai team trains local staff. After that point, passenger operations will start and the program will run for an additional two years.

A team from Greece’s Institute of Communications and Computer Science from the National Technical University Department will also support the experiment, Raptis said.

“Globally, our program is synonymous with pioneering innovation,” said Yannis Kotoulas, president of e-Trikala told TechNode. “We will be able to see how passengers and people living with the experiment react to the buses,” he said, describing the partnership with Weichai as a “huge pleasure.”

Weichai Group is a Chinese state-owned corporation that specializes in the design and manufacture of diesel engines and vehicles. It has clients in 110 countries around the world, according to its website.

“We believe not only in this particular move, but in close collaboration with them [Weichai] to take steps that the global automotive market needs,” Raptis said, referring to the bypassing of obstacles and on-demand service.

Shanghai-based DeepBlue AI was also involved in the design and manufacturing of the vehicles, people familiar with the matter told TechNode.

If it wasn’t for a DeepBlue event in Athens last June, this deal may never have gone through. Trikala Mayor Dimitris Papastergiou told TechNode that it was after this promotional event that he informed DeepBlue of the tender.

Small city, big ambitions

Primarily agricultural with little industry in the heart of Greece’s biggest valley, Trikala had fallen on hard times competing with international product prices and volumes.

Over time, it became, at best, a stop over for tourists on the way to Meteora, a UNESCO world heritage site featuring monasteries built on towering rocks reaching 550meters in height. While tourists from Russia, the Balkans, and beyond continued to flock to the important religious landmark, Trikala’s economy was dwindling.

Technology offered the city not only an opportunity to better the lives of residents but also to nurture tourism and create jobs. Tours to Meteora now stop at Trikala to see the city’s smart infrastructure and try out the free public electric vehicles.

“We need to create our own opportunities and not wait for the state,” the mayor said. He said the municipality had submitted over 1,000 applications to international institutions for technology funds.

Trikala has gained a reputation on the European stage as the country’s first smart city. The Ministry of Economics and Finance named the city Greece’s first digital city in 2004. By 2009, it was listed in the world’s top 21 smart cities worldwide by the Intelligent Community Forum, a global network of smart communities.

The local municipality has integrated several intelligent features into the city’s infrastructure, including sensors on car parking spots, smart waste management and a pilot 5G network, one of three in the country. Chinese technology has been key to at least one of these, the engineers working on the project told TechNode.

The smart waste system was designed by local engineers and manufactured in China. The system monitors key pumps in the city’s waste pipes and alerts the control room if the pumps are under stress or in need of maintenance.

Without the option to manufacture cheap and quality hardware in China, implementing the system would have been far more difficult, the engineers said.

In 2015 and 2016, Trikala ran another driverless bus pilot funded by the European Union. Among the seven cities that participated in the project, Trikala was the only one to launch the project downtown. It served well as a tourist attraction, e-Trikala President Kotoulas said.

Results from the EU study showed that passengers at Trikala were unique in using the driverless bus regularly, as opposed to just out of curiosity. This data concurs with what local authorities told TechNode. The city’s residents are used to high tech applications and are proud to be part of a community that innovates.

The municipality anticipates further collaboration with Weichai in automated and sustainable mobility in the near future.

Researchers from artificial intelligence startup Sensetime have created the “largest” benchmark for deepfake detectors, allowing developers to train and test systems that attempt to identify face forgeries.

Why it matters: Used to manipulate media using artificial intelligence to create realistic-looking videos, images, or sounds, deepfake technology has sparked concerns that applications including facial recognition systems could be fooled, leading to compromised personal data. In popular cases, celebrities’ faces have been superimposed on bodies that are not their own.

• China has enacted rules that require online platforms to clearly mark content that has been created using deepfakes, deep learning, virtual reality, or other technologies.

“The popularization of ‘deepfakes’ on the internet has further set off alarm bells among the general public and authorities, in view of the conceivable perilous implications. Accordingly, there is a dire need for countermeasures to be in place promptly, particularly innovations that can effectively detect videos that have been manipulated.”

—Researchers from Sensetime and Nanyang Technological University

China targets ‘deepfake’ content with new regulation

Details: Sensetime Research created the benchmark, dubbed Deeper-Forensics-1.0, along with Singapore’s Nanyang Technological University. The researchers claim the dataset is 10 times larger than others of its kind, consisting of 60,000 videos made up of 17.6 million frames.

• Sensetime said that all source videos were carefully collected, and fake videos were generated with a newly proposed face-swapping framework.
• The researchers collected face data from 100 paid actors from 26 countries, 53 of which were male and 47 female, to build DeeperForensics-1.0. The actors fell between the ages and 20 and 45, which is the most common age group that appears in real-world videos, Sensetime said.
• The researchers added that the videos in the dataset are “more realistic” than others that exist. They also included footage at different rates of compression and blur in order to mimic real-world scenarios.
• The benchmark also includes a hidden test set, which contains manipulated videos that were able to trick human evaluators, they said.

Context: Deepfakes gained widespread attention in China in September when popular face-swapping platform Zao was thrust into the spotlight over privacy issues. Released on August 31, the app quickly went viral in China before its policies allowing excessive data collection were publicized. Zao quickly became a sensation, with its servers hitting maximum capacity on the day of its launch.

• Regulators took notice, summoning executives from its parent company, dating platform Momo, to discuss Zao’s data collection practices.

China’s securities watchdog has approved an application from chipmaker Rockchip to list on the Shanghai bourse, as the company taps the capital markets following an unsuccessful attempt to go public three years ago.

Why it matters: Founded in 2001, Rockchip failed to list on Shenzhen’s ChiNext board in 2017 for “critical sales stagnation and asset decline,” according to China Money Network.

• ChiNext was set up as a way for predominantly high-tech firms that do not meet the requirements of Shenzhen’s main board to go public.
• China plans to domestically produce 75% of all key components, which includes chips, by 2025 amid a protracted trade war with the US.

Details: Based in the eastern Chinese city of Fuzhou, Rockchip has launched a series of artificial intelligence (AI) chips since its failed listing.

• Rockchip’s revenue in the first half of 2019 reached RMB 574 million ($83.3 million) while its net profit was RMB 66 million.
• The company’s full-year 2018 revenue grew around 2% year on year, while its profits surged 80% to reach RMB 192 million during the same period.
• China Industrial Securities is underwriting the chipmaker’s initial public offering (IPO). The company’s roadshow will be held online on Friday.
• Rockchip said its IPO will not exceed 42 million common shares.
• The company launched its first AI chip in 2018, followed by another in 2019 that combined AI and internet of things (AIoT) infrastructure, which enabled voice recognition and face detection functionality.
• Payments platform Alipay uses Rockchip’s AI chip solution in its Dragonfly facial recognition point of sales system.
• The company holds around 400 patents, which include 27 for integrated circuit designs, according to its prospectus. Manufacturers including Samsung, Sony, Huawei, Oppo, and Vivo have adopted its products, the company said.
• It is unclear whether Rockchip will list its shares on the main Shanghai exchange or the high-tech STAR Market board.

Context: Rockchip produces chips for handheld devices including tablets and smartphones, as well as TV boxes, IoT hardware, and is looking to tap the smart speaker market.

• While China has downplayed its Made in China Initiative amid trade tensions with the US, the country is setting out ambitious goals to increase its technological independence, a move that will be driven by domestic companies.

Guangzhou city government announced on Wednesday a plan to launch a zone dedicated to artificial intelligence (AI) in a bid to build a robust digital economy.

Why it matters: Zones are concentrated policy projects. This launch demonstrates Guangzhou’s resolve to encourage data sharing, the backbone of artificial intelligence applications.

• Competition between municipalities is a defining characteristic of China’s race for technological prowess on the global stage. Cities compete to attract talent and foster high-tech industry as part of the central government’s focus on upgrading industries and fostering high technology development. On a city-by-city basis, Beijing wins out on educational resources and talent, but Guangzhou could leverage its proximity to Shenzhen, Hong Kong, and Macao.
• “In developing industrial policy, we should consider the ecosystem, rather than fragmented dots in the policy paper,” Gao Feng, Director of Open Data China and Partner of Shanghai Morrow Tech, told TechNode.
• “If we look at the whole area, there are more top universities competitive in AI than Beijing. The question is how Guangzhou can release better policies to attract or encourage cooperation,” said Gao.

Details: The AI and digital economy pilot zone will include the Pazhou area, the Higher Education Mega Center, and Yuzhu area.

• Pazhou is already home to internet companies like Tencent and Guangdong Technology Financial Group. There is also a strong e-commerce presence—Alibaba, household appliance brand Gome and online discount sales company Vipshop have offices there.
• Guangzhou’s Higher Education Mega Center houses two super computers and the campuses of some of China’s top technology universities, such as Sun Yat-sen University and South China University of Technology.

INSIGHTS | Shenzhen: Built on factories but not a factory town

Context: Guangzhou is far from the only city to announce policies supporting AI.

• Ministry of Science and Technology (MoST) proposed 20 cities be given the go-ahead to build AI pilot zones before 2023, in an AI development blueprint released in August (link in Chinese). Shenzhen also has a plan devoted to AI released by MoST.
• The provinces of Guangdong and Shandong have ordered cities within their jurisdiction to embrace the idea of open government data. Guangzhou and Shenzhen already release data, such as information on daily company registration.
• But China is still behind in releasing real-time dynamic data, with local governments deterred by the cost of building technical infrastructure for data release. They may also lack confidence that data, once released, will generate interest from companies and lead directly to ventures, according to Gao.
• Another barrier to data sharing is the risk that organizations will be punished by the government if anything goes wrong. Any plan that seriously wants to encourage data sharing must “provide a space to experiment and framework for protecting personal information in data sharing,” said Gao.

Huawei announced last week changes to its organizational structure and management team, creating a fourth business group for its cloud computing and artificial intelligence (AI) divisions in a sign that the telecommunications giant is priming to ramp up its efforts in this sector.

Why it matters: The changes echo Huawei’s “Cloud Only” strategy announced (in Chinese) in April, in which the company pledged to invest more resources and funds to build a “full-stack cloud platform.”

• The world’s largest telecom equipment maker and second-largest smartphone vendor is actively opening up new business lines amid US trade sanctions and backlash.
• A focus on cloud computing puts Huawei in direct competition with the biggest Chinese players including Alibaba and Tencent, as well as global heavyweights such as Amazon and Google.

“The future of computing is a massive market worth more than two trillion US dollars. We’ll keep investing with a strategy that focuses on four key areas. We will push the boundaries of architecture, invest in processors for all scenarios, keep clear business boundaries, and build an open ecosystem.”

— Ken Hu, Huawei deputy chairman, in a company statement from September

Details: Huawei appointed Hou Jinlong president for the cloud computing and AI business group, the company confirmed to TechNode on Tuesday. Finance news outlet Yicai first reported the news, citing an internal company document. Hou had previously led the same team, but his title had been president of the business unit.

• The changes signify the cloud computing and AI divisions have become Huawei’s fourth business group following its carrier, consumer, and enterprise businesses, meaning the segment will gain more resources and support from the company, according to the report.
• The changes are a “significant transformation” to Huawei’s information and communications technology (ICT) segment and a “step further to implement Huawei’s Cloud Only strategy,” the Yicai report cited an anonymous Huawei employee as saying.

Context: Huawei Deputy Chairman Ken Hu said in September that the company would invest $1.5 billion over the following five years to attract developers to work in its cloud platform.

• Combined revenue for Huawei carrier, enterprise, and consumer business in the first half of 2019 was RMB 398.9 billion, accounting for 99.4% of its total revenue in the period.
• The company expects revenue in 2019 to jump 18% year on year and reach RMB 850 billion (around $123.3 billion).

Editor’s note: On Jan. 13, we published a story about an unsecured Megvii server that allegedly contained data from millions of public surveillance cameras in China. We believe that data privacy is an extremely important issue and one that, especially with Chinese companies, doesn’t get enough attention. As we dug into this story, it changed from one about data privacy into something different. We realized that the topic was so technical that it would be difficult for our small, independent newsroom to reach conclusions quickly, so we decided to give what we know now its own treatment while we continue to work on a fuller story.

Our mission is to inform the world about China through the lens of technology. We do that by providing neutral and accurate coverage about the companies and people that are changing the landscape of the country’s economy and society. The original story contained ambiguities that we felt did not do our mission justice.

We believe there is an important conversation about China and its companies that doesn’t happen often enough. In order to push that conversation forward, we’ve decided to replace the original text of this story with the text you are reading now. We will be writing more about this in the following weeks. Look out for our members-only weekly newsletter where we will explore this story in more detail. Sign up now so you don’t miss it.

Large-scale blockchain applications with more than 10 million daily active items could gain mass adoption in 2020, according to a report from Alibaba global research program DAMO (Discovery, Adventure, Momentum, and Outlook) Academy on Wednesday.

Why it matters: China’s tech mainstays are piling in on the global drive toward deep-technology innovations with increasing focus and investment in research and development (R&D) capabilities.

• This year’s report is the second time that Alibaba’s in-house R&D division has weighed in on major tech trends. DAMO predicted that real-time simulation of smart cities, commercialized blockchain technology, and applied 5G application would move forward in 2019.

Details: This year’s predictions cover multiple deep-tech sectors, from artificial intelligence and blockchain to the industrial internet of things.

• Large-scale production-grade blockchain applications will gain mass adoption, the report states. The blockchain-as-a-service model will further reduce barriers of entry for enterprise applications.
• AI will evolve from perceptual intelligence to cognitive intelligence, drawing inspiration from cognitive psychology, neuroscience, and human social history, the report continued.
• The report predicts that processing-in-memory (PIM) innovations will spur on next-generation AI. PIM architecture involves the fusing of memory and processor together. Computations take place where data is stored with minimal data movement, allowing improved computation parallelism and efficiency.
• DAMO expects the onset of 5G, the rapid development of IoT devices, as well as cloud and edge computing, to accelerate the fusion of information, communications, and industrial control systems in 2020.
• Large-scale collaboration between machines will also become possible thanks to collaborations involving IoT and 5G, the report states.
• The report also expects modular design to ease chip manufacturing and new materials to revolutionize semiconductor devices.

Context: Alibaba Group launched DAMO Academy in October 2017 to boost technical collaboration worldwide.

• The $15 billion high tech unit has offices in Beijing, Hangzhou, Singapore, Seattle, Sunnyvale, Tel-Aviv, and New York.
• DAMO has become an R&D engine for the tech giant, backing a series of projects from its neural network chip to cloud computing.
• In addition to DAMO Academy, Alibaba leverages Luohan Academy to boost open-research.

Alibaba’s Luohan Academy celebrates first anniversary

If you can’t see the YouTube player above, try watching here instead.

Baidu has released its annual ranking of the hottest search terms in technology for 2019.

Artificial intelligence (AI) garnered more searches than any other tech phrase.

“AI is going to open a new chapter of the society of the world that people try to understand ourselves better, rather than the outside world,” said Alibaba founder Jack Ma in a discussion with Tesla CEO Elon Musk at the World Artificial Intelligence Conference in Shanghai in August.

China issued a plan for next-generation AI in 2017, pledging to turn the industry into a new growth engine. Countless Chinese entrepreneurs and startups have focused on AI in recent years.

Hurun named SenseTime and Megvii in the top four of its Hurun Global Unicorn List 2019 with respective valuations of USD 6 billion and USD 4 billion.

5G was the second most searched term

It was a pivotal year for the next-generation communications technology in China as the country officially kicked off commercialization services in November, giving it a slight lead in the global race to build the superfast networks.

For consumers, faster speeds and low latency can improve mobile experiences. For industry, machine-to-machine communication on a massive scale presents vast potential for industrial IoT.

A report from the China Academy of Information and Communications Technology estimated that the domestic 5G market could be worth RMB 1.1 trillion by 2025, contributing 3.2% of China’s gross national product expansion.

Blockchain ranked third

China has stepped up efforts to apply blockchain in a wide range of fields, with President Xi pointing out the importance of the technology in a landmark speech in October.

China leads global blockchain patent applications, according to a report from the China Academy for Information and Communications Technology. Alibaba, Ping An, Baidu are spearheading the development of China’s blockchain sector.

The country’s central bank is also developing its own digital currency, which has created a climate of confidence for the already fast-moving blockchain sector. Spending is likely to hit $2 billion by 2023, according to a report from global market intelligence company IDC.

VR: 5, AR: 10

2019 also witnessed the development of virtual reality (VR) and augmented reality (AR) which also made Baidu’s top 10.

China’s VR market will grow to RMB 54.5 billion ( USD 7.7 billion) by 2021, according to the Ministry of Industry and Information Technology.

AR is becoming more and more popular in the country. Scanning objects for virtual red envelopes using Alipay has already become a widespread practice during Chinese New Year. 5G is expected to aid the technology’s development further.

Below is Baidu’s top ten in full.

1.     AI
2.     5G
3.     Blockchain
4.     Robotics
5.     VR
6.     AI for missing people
7.     Smart Home
8.     IoT
9.     Facial Recognition Payment
10.    AR

China became the largest market globally for smart speakers earlier this year with some 10.6 million units shipped in the first quarter. In fact, three of the world’s five leading smart speaker vendors in the third quarter were Chinese, according to market research firm Canalys. 

The top Chinese players are household names: e-commerce giant Alibaba, search operator Baidu, and handset maker Xiaomi. They all rely on a little-known startup called SoundAI for critical voice interaction technology.

The backstory: The Beijing-based startup makes voice recognition and artificial intelligence (AI) software, helping smart speakers from leading manufacturers to listen to and process users’ requests.

• The company’s products include Azero, a voice interaction kit that runs on smart speakers, connected cars, and wearables; Babel, a voice recognition software; and Cimon, an audio processing tool.
• SoundAI raised RMB 200 million ($28.6 million) in its most recent funding round (Series B) in December 2018. It valued the company at RMB 1 billion (around $143 million).
• Chen Xiaoliang, a researcher at the Institute of Acoustics under the Chinese Academy of Sciences, founded the firm in May 2016.

Unique selling point: SoundAI’s technology is found in more than 20 million products, ranging from smart speakers and conference systems to robots and connected cars. It also runs a strong research and development (R&D) arm with more than 1,000 patents secured so far, according to its website.

“Voice is the most natural way of communication, and smart speakers will see great demand in the near future. Our voice technology has been used in the smart speaker offerings from top players including Baidu, Alibaba, Tencent, Huawei, and Xiaomi. Our cooperation with these firms is not easy to replace.”

—Chen Xiaoliang, founder and CEO of SoundAI, in an interview with TechNode

The investors: The company has closed four rounds of investment to date, bringing in investors such as Baidu, FreesFund, Qihoo 360, Aplus Capital, and the Bank of Beijing.

• Qihoo 360, better known as a cybersecurity firm, and Baidu both make smart speakers. Baidu was the third-largest smart speaker vendor globally in the third quarter.

Present condition: The company has a team of around 200 employees, mainly from top Chinese universities such as Tsinghua and Peking University, as well as tech companies including Google, Broadcom, Tencent, and Baidu.

• The company works with 150 companies in sectors ranging from smart home and education to healthcare and manufacturing, according to Chen.
• Current information on the company’s profitability was unavailable at the time of writing.

The landscape: Global smart speaker shipments grew by more than half year on year to hit 34.9 million units in the third quarter, according to market research firm Strategy Analytics. China accounted for 36% of global shipments.

• In the two years since their emergence in the Chinese market, smart speakers have evolved from niche gadgets into one of the most sought after devices in Chinese households.
• Smart speakers are a key item for controlling intelligent home platforms, a booming sector.
• The device’s popularity in China is partly due to a price war waged between the country’s tech heavyweights, with average prices slashed to under $20 from about $100 in 2017.
• Another Strategy Analytics report predicts global consumer spending on smart home-related hardware, services, and installations to reach $103 billion in 2019 and $157 billion in 2023.
• As a voice recognition technology provider, SoundAI shares the market with smart home solution provider Tuya Smart and Unisound, a Qualcomm-backed voice-to-text software maker, among others.

China’s tech giants battle for smart speaker supremacy as price war rages on

Prospects: The company is likely to maintain growth as the smart speaker becomes a fixture in the “vast majority” of Chinese households. Baidu’s recent move to pull out of the price war by cutting subsidies indicates organic demand is increasing.

• SoundAI is on a list of 100 tech companies that Chinese business news platform Sina Finance expects to go public on the STAR Market, the country’s Nasdaq-style tech board on the Shanghai Stock Exchange.

Data security startup DataCloak is using artificial intelligence (AI) and “zero-trust” computing to give companies greater control over their sensitive information to combat internal and external threats.

The backstory: Founded in early 2018 by a former senior director and engineers from search giant Baidu, DataCloak provides solutions to ensure sensitive data remains confidential.

• The Shenzhen-based company comprises former Google, Tencent and Baidu employees, with CEO Liu Chao setting up the company alongside co-founders Wu Ye and Yang Yifei.
• The company has already raised $5 million and is seeking additional funding despite its recent formation.

Unique selling point: DataCloak wants to allow employees of an enterprise to securely access corporate data wherever they are in the world. They do this by combining multiple technologies that allow for increased control over who can access privileged information.

• DataCloak follows the “zero-trust” approach to data security, where users are not trusted purely because they are on a corporate network.
• The company achieves this by setting up an isolated and secured workspace within an employee’s computer or smartphone, with only registered devices able to access a corporate network.
• While files and clipboard data can flow freely into the workspace, information flowing from within can be controlled and later audited, lowering risks of data leaks.
• DataCloak provides secure communication between a device and a corporate network from anywhere in the world, and allows for configurable security levels for every device. This permits a company to choose who can access what data.
• The platform also features AI-driven security, which learns a workspace users’ normal behavior to identify abnormalities that could act as an alert for a potential threat.

“We try to protect the owners of data during the whole data lifecycle, from generation to transformation, as well as the data flow. This could be the source code from a software company, it could be users’ private data held by banks, or even invoices or customer lists. We don’t trust based on where you are; we don’t trust you based on our account name; we don’t trust a laptop purely because it’s distributed by your company. We use a ‘don’t trust, always verify’ approach”

—Cui Yongxin, chief operating officer of DataCloak 

The investors: DataCloak has attracted attention from Matrix Partners China, which led its $5 million Pre-A Series in September 2018. The company is currently finalizing its Series A.

The landscape: China’s cybersecurity industry will reach RMB 63 billion ($8.95 billion) this year, according to figures from the China Academy of Information and Communications Technology.

• Data leaks are a common occurrence in China despite strict laws to limit the fallout.
• Enterprises that do not sufficiently protect their customers’ data can face severe penalties, including having their business licenses revoked.
• Numerous high-profile tech companies in China have suffered data breaches. In April, drone maker DJI said an employee leak cost the company RMB 1.1 million.
• Anime-streaming platform Bilibili has also had its share of problems, with user data shared in a Github repository earlier this year.
• In some cases, stolen data has gone for as little as $0.01.

Prospects: China’s 2017 Cybersecurity Law and an upcoming law governing cryptography have forced companies to improve their security.

• This has created a growing market for cybersecurity companies looking to help other firms with compliance.
• Additionally, as more employees bring their own devices to work, increasingly innovative solutions are needed to make sure corporate data remains safe.
• Data brings both wealth and power to those who hold it, increasing the need to ensure that it is kept secure and that access to critical data is controlled.

With contributions from Eliza Gkritsi

Amidst a sea of startups on show at TechCrunch Shenzhen 2019, two startups stood out for innovating in a massive industry that has remained unchanged for decades—industrial maintenance. 

Founded in 2016 by two Canadians living in Shenzhen, Skygauge has developed a uniquely stable and precise drone that tilts four rotors to keep the main body stable, even in strong winds. It is equipped with a sensor that attaches to a metal structure, such as an oil rig, and collects corrosion data within a few seconds. 

“Our frame can remain perfectly level, so that we can precisely approach a structure and contact the structure in very specific locations,” Nikita Iliushkin, co-founder and CEO of the startup, told TechNode. 

Thanks to the stability, the Skygauge drone can conduct inspections of large metal structures, like oil and gas facilities, offshore platforms, bridges, and ships. Currently, such checks usually require scaffolding around the structure to make it accessible—a task that can take weeks. Skygauge can not only save businesses time and money but also put workers out of harm’s way, Illiushkin said.

Skygauge sees inspections as only the first application of its stable drone technology.  In the future, the team wants to attach a stable robotic arm for general work, so that the drone can carry out jobs like painting, washing, and coating.

 “The same way robotic arms revolutionized work in factories, our drones will revolutionize work in the skies,” Iliushkin said. 

Also based in Shenzhen, Cassandra has developed an IoT-based predictive maintenance solution. The solution involves attaching magnetic ultrasonic vibration sensors to industrial machinery, and wirelessly feeding data in an artificially intelligent algorithm. Using Cassandra’s system, factory operators can monitor in real-time variables such as temperature, speed, and corrosion on an iPad app.

The AI analyzes the data and provides predictions on the machine’s state and functionality, so that the operators can know when a machine will fail ahead of time. 

“86% of all maintenance is either scheduled unnecessarily or too late [currently],” said Alain Garner, founder and architecture director at Cassandra, a Cogobuy Group company. “Either people are using time-based maintenance, so just throwing away things before they break,” he continued. 

Cassandra’s solution can help businesses save money on unnecessary maintenance or downtime caused by maintenance operations. “We’re in deployment at the moment in pilot stages, and we’re sort of rolling this out further,” Fournier said. 

The predictive maintenance market is expected to grow to $10.7 billion by 2024, according to global research firm MarketsandMarkets.

Artificial intelligence (AI) firm Megvii has delayed its Hong Kong listing until next year following additional queries from the stock exchange as it prepares to go public, Nikkei Asian Review reported.

Why it matters: Megvii was one of several Chinese AI companies added to the American so-called Entity List in October for alleged complicity in human rights violations in China.

• The world’s most valuable AI startup Sensetime, speech recognition firm iFlytek, surveillance equipment manufacturer Hikvision, and AI company Yitu were also included in the blacklisting.
• The ban has investors worried as it effectively bars the companies from doing business with American firms, cutting them off from purchasing US-made components.

Details: Megvii had initially timed its initial public offering (IPO) to take place before the end of the year. However, the company decided to push the date to next year after additional scrutiny from the Hong Kong stock exchange, according to separate reports from Nikkei Asian Review and International Financing Review.

• The company was questioned about the adequacy of its risk disclosures related to US sanctions in its IPO filing.
• Megvii filed for a Hong Kong listing in August but was mulling a delay prior to queries from the bourse.

Context: Megvii was expected to become the first of China’s AI startups to go public. The company would have acted as a litmus test of sorts, paving the way for other companies in the industry.

• Megvii has also faced scrutiny in Taiwan, where the government planned to investigate the company following the ban after it was awarded a contract to install a security system at Taichung Power Plant.

Bottom line: Potential investors are hesitant about buying shares in Megvii after the blacklisting. The company may have trouble hitting a $3.5 billion valuation, a source previously told Bloomberg. The company was valued at $4 billion in its latest funding round.

Chinese authorities rolled out a new regulatory document on Friday targeting so-called “deepfake” images and video with rules calling on media platforms to identify and remove files spreading false news.

Why it matters: Online video and audio platforms are under more pressure to review content. Standards detailing deepfakes, or media in which figures in images or videos are swapped with another person’s likeness, will likely follow on the heels of the document.

Details: Three government agencies—the Cyberspace Administration of China (CAC), the Ministry of Culture and Tourism, and the National Radio and Television Administration—released the document, calling on platforms to more clearly mark audio or videos using deepfakes, deep learning, virtual reality or other new technologies.

• The document, effective Jan. 1, bans the use of deepfake and virtual reality (VR) technologies in creating, publishing or spreading fake news, and calls on platforms to remove such media.
• Users must register on platforms with identifiable information like government-issued IDs or mobile phone numbers, in line with the Cybersecurity Law.
• Platforms should set up easy-to-use complaint channels.
• Audio and visual services should issue industry standards and guidelines, and set up a credit system.
• China already has 759 million online video platform users, according to an unnamed CAC representative cited in an accompanying document. New technologies such as deepfakes could “endanger national security, disrupt social stability, disrupt social order, and infringe on legitimate rights and interests of others,” the document said.
• Government departments must organize regular inspections to ensure that platforms regulate online audio and video in line with service agreements.

“Currently China is not facing any serious problems with deepfakes. But the threshold for this technology is getting lower and fakes are increasing in sophistication. There is no guarantee that this technology will not be abused. If abused, it can cause serious social problems and security risks” (our translation).

—Jing Dong, associate researcher at the Institute of Automation, Chinese Academy of Sciences

Context: While false news is widespread in China, deepfakes are still relatively rare.

• An employee of a video-streaming company told TechNode that scammers are doing just fine without deepfakes. Platforms already battle fake celebrity profiles set up to swindle money from fans, the person said. Fraudsters have even made fake versions of paid-streaming websites to con fans who believe they are recharging credits to send virtual gifts but are instead sending funds straight to thieves’ pockets.
• “Using tech to detect deepfakes will always be an arms race,” a Europe-based artificial intelligence researcher told TechNode in an email. “Systematically marking synthetic content could help to have a good training database for recognizing such content in the wild.”
• While doubtful about how useful real-name registration can be, the researcher added that “complaint channels are quite useful to detect instances of deepfakes due to human knowledge about contexts and situations but are also expensive to set up and run, and can be swamped.”

People in China will need to undergo a facial recognition scan when buying new SIM cards, according to rules introduced on Sunday, as the country seeks to tackle telecommunications fraud and improve cybersecurity.

Why it matters: Facial recognition is ubiquitous in China, with applications ranging from payments to public security.

• Nevertheless, the technology’s increased presence has been met with pushback as some question whether it is being overused.
• In November, a university professor from Zhejiang province in eastern China filed a lawsuit against a wildlife park which required using its facial recognition system to access the facility.
• The measure further limits online anonymity under China’s real-name verification system by implementing additional identification checks.

Details: The rule ensures that internet personas are tied to real identities as online platforms typically require users to register their phone numbers when signing up for services requiring real-name verification.

• China’s Ministry of Industry and Information Technology (MIIT) announced the rule in September, requiring enforcement of the new measure on Dec. 1.
• The move aims to prevent SIM card resales and to crack down on telecommunication fraud, as well as improve cybersecurity and anti-terrorism campaigns, according to MIIT.
• Previously, prospective mobile service buyers were only required to present their identity cards when registering for new SIM cards.

Sensetime-led consortium to set up standards for facial recognition tech

Context: China last month set up a working group for facial recognition standards that aims to assuage concerns over data security issues surrounding the technology.

• The group is led by artificial intelligence startup Sensetime, which the US has blacklisted along with several others over their alleged complicity in human rights violations in China.
• Sensetime said in a statement that the group would also participate in international standard-setting.
• Meanwhile, companies including surveillance equipment manufacturer Dahua Technology, China Telecom, and ZTE have been proposing new international standards for facial recognition, video monitoring, and surveillance at the United Nation, the Financial Times reported.

Alibaba Cloud, a subsidiary of Chinese e-commerce giant Alibaba, has opened to the public its source code for an in-house machine-learning platform that it used to drive product recommendations during this year’s Singles Day shopping festival.

Why it matters: Alibaba has sharpened its focus on open-source software since 2011. The company’s cloud division is a member of the Linux Foundation and is active in a number of open-source communities including the Apache Software Foundation.

• Against the backdrop of the Sino-US trade war, China has been criticized for relying on American frameworks for artificial intelligence (AI) and machine learning.
• AI and machine learning is a collaborative movement and Alibaba Cloud joins a number of global internet firms which have released open-source platforms. It also stands to benefit from allowing more people to work on the platform, potentially improving it.

Details: Dubbed Alink, the platform offers a range of algorithm libraries that allow for processing live data as well as batched datasets, Alibaba Cloud said in a statement on Thursday.

• Alink facilitates tasks including statistical analysis, machine learning, real-time prediction, personalized recommendations, and anomaly detection.
• Alibaba has been using the platform across its businesses for the last three years. It said Alink helped increase the click rate of product recommendations on online marketplace Tmall by 4% during this year’s Nov. 11 Singles Day shopping festival.
• The platform is built on Apache Flink, an open-source framework for handling big data at large scale.

“The difference between Alink and the pure AI platforms like Tensorflow and PyTorch is that those focus more on the algorithms and the design of the models. But today, for machine learning models to train effectively, we need to have a high-quality connection to the big data. Alink provides us with a seamless connection between the AI algorithms and the big data distributed systems.”

—Jia Yangqing, president and senior fellow of Data Platform at Alibaba Cloud Intelligence to TechNode on Thursday

Context: The State Council, China’s cabinet, has set ambitious goals for the country to become a global leader in AI by 2030. As a result, China has applied AI across industries, permeating every facet of daily life.

• Nevertheless, the country lags behind in terms of talent and semiconductor development—the tools that provide the computing power that enables machines to learn from vast quantities of data.
• Alibaba is not the only Chinese tech giant to open source its machine learning tools. Baidu has opened up its Paddle Paddle deep-learning framework, a tool that was also used internally across its product offering.

Artificial intelligence (AI) firm Megvii plans to seek approval for its Hong Kong listing on Thursday, aiming to raise at least $500 million, Reuters reported citing people familiar with the matter.

Why it matters: In October, the US government added Megvii to the so-called “Entity List” for alleged complicity in Beijing’s human rights abuses in China. The move effectively blocks the company from sourcing American-made components for its products.

• Megvii has denied the allegation and said its inclusion on the list comes as a result of a “misunderstanding.”
• The company’s initial public offering (IPO) has been delayed as a result of the US-China trade war and blacklisting.

Details: Megvii was previously seeking a fourth-quarter listing of up to $1 billion, according to Reuters.

• The company has yet to decide whether to carry out a roadshow once the application is approved, the people said.
• Goldman Sachs said after Megvii’s blacklisting that it was re-evaluating its involvement in the IPO given “recent developments.”
• The company filed for a Hong Kong IPO in August but gave no indication of how much it was seeking to raise.

Context: Megvii is one of a handful of startups that is spearheading China’s AI boom, including Sensetime, Yitu, and Cloudwalk, though none have yet gone public.

• These companies have benefited from broad government goals to build China into an AI front-runner by 2030.
• The company earns nearly three quarters of its sales from AI services for government agencies and the healthcare industry, among others.
• In May, Megvii raised $750 million in a round led by Bank of China’s equity arm, valuing the company at more than $4 billion.

Biometrics firm Megvii’s contract in Taiwan at risk after US blacklisting

In civil aviation, “innovation is still incremental and not disruptive,” but market entry in aerospace has been disrupted, said Philipp Visotschnig, CFO of Airbus China said at TechCrunch Shenzhen 2019.

The space industry has been disrupted and private access to space is possible, which was long considered impossible, he said. Elon Musk’s SpaceX approached rockets differently in terms of vertical integration and managed to use software to reuse the rocket, he said.

But commercial and cargo airplanes remain relatively unchanged in the era of innovation. “The cycle of aviation is relatively low in aerospace” as it takes around 7 years to develop an aircraft and typically fuel consumption must be reduced by 15% for a new model, he said.

Startups usually deal with simpler and shorter regulatory approvals and so far innovation has happened around customer experience and aircraft connectivity, he said. Interesting projects are developing around 5G and connectivity, as well as digital tools that help companies interface with passengers, Visotschnig said.

A trend in aviation is using digitalization and AI  to analyze big data related to aircraft maintenance, he said. “We need to make sure that it [the aircraft] is flying for the next thirty years. Having all this data is helpful,” he said. Logistics and fleet support are two areas where Airbus is looking for partnerships.

The industry needs to adopt a “far more open” approach in order to integrate new technologies into its offerings, he said. Airbus is partnering with startups at its Innovation Centre in Shenzhen, opened in 2017, where it is developing vertical lift, battery, electrification, and customer experience solutions, Visotschnig said.

Airbus is also setting up a model to commercialise flying taxis in China in two to three years, he said. The European aviation company has launched a fleet of helicopters to hire in Sao Paulo in Brazil, through its local subsidiary Voom.

A US congressional advisory body has warned that China’s focus on developing artificial intelligence (AI) could have marked effects on the global economic and military balance, shifting the seat of power from America to China.

Why it matters: Beijing has set ambitious goals to become a world leader in AI by 2030. Sensetime, the world’s most valuable AI startup, comes from China.

• Nevertheless, the country faces a slew of challenges in reaching this goal. While China’s internet population generates gargantuan amounts of data that can be used to train an AI, its domestic chipmaking sector, which the AI industry relies on, is largely inferior.
• Likewise, China is known for quickly implementing AI applications but lacks talent compared with the US.

“Chinese firms and research institutes are advancing uses of AI that could undermine US economic leadership and provide an asymmetrical advantage in warfare.”

—US-China Economic and Security Review Commission 

Details: China is prioritizing AI as it underpins the development of many other technologies, and could lead to “substantial scientific breakthroughs, economic disruption, enduring economic breakthroughs, and rapid changes in military capabilities,” the US-China Economic and Security Review Commission said in its 2019 report released on Thursday.

• The commission was set up in 2000 to evaluate the national security implications of economic ties between the US and China.
• China’s military strategists see AI as a “breakout technology” that could develop tactics to exploit US vulnerabilities, the report said.
• China’s policies on civil-military fusion seeking to leverage private sector innovation for the defense sector are worrisome for the US given the breadth and opacity of the campaign, the commission warned.
• Washington is concerned that the close collaboration between China’s military and private sector, which in turn works with US companies, could give China a leg up in a global arms race.
• The body highlighted China’s advantages in manufacturing, which could allow the country beat out the US in commercializing mass market and military discoveries made and funded in America.

Context: In the midst of the protracted US-China trade war, several high-profile Chinese AI companies have found themselves in the crosshairs.

• In October, the US barred Sensetime, as well as surveillance equipment manufacturer Hikvision and AI firms iFlytek and Yitu, among others, from doing business with American firms, effectively cutting off their supply of US-made components.
• Meanwhile, US-based think tank, Center for Data Innovation, has warned that the close ties between China’s military and private sector could hurt the country’s goals of becoming a world AI leader as distrust of companies linked to China’s government grows against the backdrop of US-China trade tensions.

China’s ‘military-civil’ partnerships could hurt its AI ambitions: report

“China’s artificial intelligence (AI) is like a kid with an IQ of 120. The US’ is a child with an IQ of 140. But in the US, the kid hasn’t left the laboratory,” according to Zhou Wei, founder and managing partner of China Creation Ventures.

Zhou said that China’s AI industry could overtake that of the US given the country’s move toward mass implementation of the technology and the fast rates of business model iteration.

“In the long run, the Chinese kid could outpace the US child,” he said during a fireside chat at TechCrunch Shenzhen 2019 on Monday.

Zhou’s comments come in the midst of a protracted US-China trade war, in which several high profile Chinese AI companies have been caught in the crosshairs.

The world’s most valuable AI startup Sensetime, as well as surveillance equipment maker Hikvision and speech recognition firm iFlytek, among others, were placed on a US trade blacklist last month for their alleged complicity in human rights violations in China.

Zhou’s sentiments echo those of Chinese AI expert and founder of Sinovation Ventures, Kai-fu Lee, who claimed previously that China mass-implementation of AI, access to data, and size of China’s internet population could put the country at the forefront of the technology’s development.

“Business models in China are iterated quickly, while Silicon Valley entrepreneurs and investors want to come up with a product that is perfect and doesn’t have too many problems,” Zhou said.

AI applications permeate every aspect of life in China, where it used from everything from facilitating facial recognition payments to keeping tabs on the country’s population and assessing potential risks related to loans.

The country has also become a haven for data production as internet users’ daily lives move online. For example, mobility services including bike-sharing and ride-hailing generate gargantuan amounts of data that can be used for training an AI. China’s biggest ride-hailing platform Didi recently released a transit dataset to help researchers in a push to better understand transport patterns and optimize infrastructure investments.

Nevertheless, China faces the prospect of a major brain drain, according to MacroPolo, a China-focused think tank at the Paulson Institute in Chicago. Around three-quarters of the country’s AI talent is currently located outside of China, the research body found for its analysis of submissions to the NeurIPS conference.

Speed is a requirement for companies in China, said Zhou, adding that whoever is fastest will be the winner. “China’s consumers are tolerant of new projects and also impatient,” he said.

Zhou noted that this fast aspect has also become a problem for venture capital firms, who can struggle to keep up. “We’re really under a lot of pressure to learn quickly and to be very focused,” he added.

Search giant Baidu beat analyst expectations for its third quarter revenues as the company’s diversification away from its core search business showed signs of paying off.

Why it matters: Baidu has seen increased competition for advertising revenue from rivals including Bytedance and Tencent in the midst of a macroeconomic slowdown that has led advertisers to tighten their belts.

• Baidu has had a tough year—the company’s share price fell by more than 35% prior to its latest earnings release.
• The search giant this year posted its first quarterly loss since listing in 2005.
• Baidu CEO Robin Li warned in January that “winter is coming,” acknowledging the effects of China’s slowing economy and the fallout from the trade war with the US.
• The company has since implemented numerous cost-cutting measures to mitigate risks to its business.

Details: Baidu’s Q3 revenue reached RMB 28.1 billion (around $4 billion), beating analyst expectations of RMB 27.5 billion. Revenue was up 7% compared with the second quarter.

• The company posted a net loss of RMB 6.4 billion. Baidu made a profit of RMB 12.4 billion during the same period last year.
• The company, in part, attributed the losses to its investment in online travel agency Trip.com, whose share price has declined, according to Baidu. In October, Baidu reduced its holdings in Trip.com, selling $1 billion of its shares.
• Q3 revenue from Baidu’s video-streaming platform iQiyi increased 7% compared with the same quarter a year ago, reaching RMB 7.4 billion, boosted by 31% year on year growth in subscribers to nearly 106 million.
• While iQiyi’s subscription revenue increased, the platform’s advertising sales shrunk as a result of increased competition, Herman Yu, Baidu’s chief financial officer, said during the company’s earnings call.
• Meanwhile, online marketing revenue fell by 9% compared with the same period last year as competition and the slowing economy took their toll.
• Baidu expects revenues of between RMB 27.1 and RMB 28.7 billion in the fourth quarter, representing growth of -1% to 7%.
• Baidu’s share price rose 4% in after-hours trading on Wednesday.

Context: Baidu has plowed billions into diversifying its offerings, particularly on artificial intelligence and cloud computing, and is looking to enterprise services for growth.

• Nonetheless, several of these investments, most notably its bet on autonomous driving, have yet to pay off. It’s likely that the company’s focus on self-driving cars will only begin to bear fruit in the next five to 10 years, putting additional pressure on its other businesses.

Artificial intelligence (AI) startup Megvii is considering whether to delay its Hong Kong listing despite plans to go public before year-end, Bloomberg reported.

Why it matters: Megvii was among several of China’s biggest AI firms added to the US government’s so-called Entity List in October, effectively blocking them from sourcing American-made components.

• The move has raised questions about the future of these companies, as some source components from US manufacturers, including Nvidia.
• Following Megvii’s blacklisting, Goldman Sacks, one of the IPO underwriters, said it was reevaluating its involvement in the company going public.
• Taiwan’s government plans to investigate the firm following the ban, after Megvii was awarded a contract to install a security system for the Taichung Power Plant.

Undeterred by US blacklisting, AI firm Megvii eyes end-year IPO

Details: Megvii is currently discussing with advisers whether to press ahead with the listing this month or postpone until the company is removed from the Entity List, sources told Bloomberg.

• Investors are worried about buying shares and Megvii may have trouble hitting a valuation of $3.5 billion, one of the people said. The company was previously valued at $4 billion after its last round of funding.
• The company still plans to hold its listing hearing in Hong Kong this month.
• Megvii said previously that it believes its inclusion on the blacklist was a “misunderstanding,” adding that the company would be “engaging with the US government on this basis.”

Context: Megvii was expected to become China’s first AI startup to go public, acting as a possible litmus test and paving the way for other companies in the industry to follow.

• The US blacklisting came after Megivii, along with the world’s most valuable AI startup Sensetime and surveillance camera manufacturer Hikvision, were accused of complicity in human rights violations against Muslim minority groups in China.
• Speech recognition giant iFlytek, as well rival AI startup Yitu and surveillance firm Dahua Technology, among others, were also added to the list.

Artificial intelligence (AI) startup Clobotics is looking to cash in on China’s massive wind energy boom, using its technology to automate turbine inspections that typically take up to six hours when done manually.

The backstory: Founded in 2016, Clobotics provides computer vision and data analytics solutions for the wind power and retail industries. The company aims to automate time-intensive processes through data analysis.

• With headquarters in Shanghai and Seattle, Clobotics has partnered with companies including Shanghai Electric, LM Wind Power, Coca Cola, Concord New Energy, and GEV Wind Power.
• The company was founded by executives from Chinese drone maker Ehang, including its former COO George Yan and Claire Chen, previously general manager of the company’s North American business.

Unique selling point: Clobotics claims to be able to complete inspections of wind turbines in 25 minutes, with the help of autonomous drones and its computer vision platform, a process that typically takes six hours when done manually. This could dramatically reduce labor costs and the amount of time a turbine is out of service during the inspection process.

• Clobotics’ computer vision module can detect defects as small as 1 millimeter by 3 millimeters, the company says.
• After an inspection is completed, images are uploaded to the cloud where the company’s analytics platform examines the imagery and identifies issues with the blades.
• The company has forged several major partnerships with key players in the power generation industry.
• Given the height of the turbines, automated solutions could dramatically decrease risks during inspections.

“We are committed to providing the best AI solutions to digitalize the traditional industries. When smart devices marry computer vision technologies, massive amounts of offline data are collected, recognized, and analyzed automatically through artificial intelligence.”

—Clobotics co-founder George Yan

The investors: Clobotics has attracted investors including GGV Capital, CDIB Capital, the overseas-focused investment arm of Taiwan-based China Development Financial, and CMC Capital. It most recently closed its $22 million Series Pre-B.

• The company has secured a total of $43 million in funding, according to data from Crunchbase.

Present condition: Clobotics is currently seeking to increase its market share in the US and Europe. The company this year partnered with GEV Wind Power, a company providing wind turbine maintenance services globally, expanding the Chinese startup’s reach to North America, Europe, and Africa.

• The company said shortly after its latest funding round that it would continue to seek growth in the US and European markets.
• Nevertheless, leaders in the wind industry have warned that the protracted US-China trade war could hurt the development of the industry as a whole, as insiders fear that the cost of wind turbines could increase due to protectionist policies.
• Meanwhile, the Chinese government has announced it will end subsidies for new onshore wind power projects by 2021.

The landscape: Drones have been used in inspections in the past, but Clobotics’ system includes autonomous drones that automate the process.

• The company currently has few homegrown competitors, as the majority of the inspections are done manually. Nevertheless, a US-based rival has moved into China through a domestic tie-up.
• California-based Sterblue has partnered with Chinese turbine inspection firm Winward to introduce its AI-based drone inspection platform to China.
• The market is huge and is set to grow further. China installed nearly 20 gigawatts of capacity in 2017 alone and is home to the biggest onshore wind farm in its western province of Gansu.
• The country will reach nearly 440 gigawatts of capacity by 2028, with onshore farms accounting for 84% of all new developments, research firm Wood Mackenzie said in a report from August.
• Around 410 wind turbines are needed to produce 1 gigawatt of power, according to estimates by the US’ Office of Energy Efficiency and Renewable Energy.

Prospects: Clobotics has in the past year seen increasing attention from new and old investors, closing two rounds of funding collectively worth $33 million. As developers rush to install new capacity before China drops subsidies in 2021, Clobotics could see a drastic increase in the adoption of its services.

The company has also not limited itself to the Chinese market, as it seeks to expand to offer its autonomous inspections globally and take advantage of the renewable energy boom that has arisen over escalating concerns over carbon emissions.

Led by ex-Google chief Eric Schmidt, a US defense committee urged the government to increase its spending on artificial intelligence (AI) research and continue the use of export bans to ensure its lead over China.

Why it matters: The report is the latest and most comprehensive effort from the Department of Defense (DoD) to create a blueprint for safeguarding American interests against the rise of China as a global leader in AI. The National Security Commission on AI (NSCAI) released its first interim report on Monday, outlining key strategic concerns and proposing next steps.

“We are a pro-America Commission, and the final report will say how we will win this competition.”

⁠—Eric Schmidt, NSCAI chair

Details: The NSCAI recommended that the US enhance domestic and allied cooperation while expanding export controls on target technologies and protecting American research from  “state-directed espionage.”

• “US and allied AI hardware advantages,” referring to the leadership in semiconductors held by the US, Japan, and the Netherlands, must be protected through multilateral export restrictions, the commission wrote.
• Due to AI’s multiple applications, traditional item-based controls may not suffice, the report said. Instead, regulators should scrutinize potential end-use and the “end user of specific items, to prevent their use for malicious purposes,” the report said.
• But the US and its allies should be careful about possible retribution from Beijing, NSCAI warned.
• Universities should be “part of the solution,” the report said, meaning that they should work with law enforcement and intelligence services to keep American secrets. Preventing “direct or indirect assistance to China’s military and intelligence apparatus” should be top priority, the commission wrote.
• At the same time, the US should be open to collaboration with China on global standard-setting for safe use of AI.
• The report named the following concerns: China’s spending on research and development of AI, the ballooning commercial competitiveness of Chinese tech giants which is “is being harnessed to promote national objectives in AI,” China’s use of AI in the military, and its transformation into a pole of attraction for global talent.

China’s share of global AI investment shrinks as headwinds take their toll

Context: The NCSAI was established by the National Defense Authorization Act of 2019, an annual US regulation that sets the defense budget. It first convened on March 11.

• China’s spending AI research and development (R & D) grew by 300% from 1991 to 2015, according to R&D Magazine. In 2017, China briefly surpassed the US in AI-focused venture capital investments. It still lags the US, but is on track to surpass it by 2030, the year when Beijing has pledged to reach global AI supremacy, according to the publication.
• In terms of academic research, China’s global impact has increased but is still behind the US. Researchers at the Allen Institute, a US think tank focused on AI, predicted that Chinese researchers will overtake their American counterparts in terms of citations by 2025.
• Washington placed several prominent Chinese AI companies under an export ban in October 2019, including Sensetime, the most valuable AI company in the world.
• American universities are increasingly scrutinizing their relationships with Chinese companies and research institutions. The Massachusetts Institute of Technology and Stanford cut their funding ties with Huawei following its inclusion on a US trade blacklist.

China is attempting to predict potential crimes using artificial intelligence (AI), employing the technology to monitor the emotional state of its citizens, the Financial Times reported.

Why it matters: China is the world’s largest surveillance market and is home to some of the world’s biggest equipment makers.

• Several prominent Chinese AI firms were last month added to a US trade blacklist for their alleged complicity in Beijing’s human rights violations in the northwestern Xinjiang Uyghur Autonomous Region.
• The blacklisting comprised a number of the country’s “AI Champions,” including Sensetime and Yitu, speech recognition firm iFlytek, and surveillance camera maker Hikvision.

“Using video footage, emotion recognition can rapidly identify criminal suspects by analyzing their mental state… to prevent illegal acts including terrorism and smuggling.”

—Li Xiaoyu, policing expert and party cadre from Xinjiang, cited by FT

Details: Emotion recognition was a hot topic at this year’s China Public Security Expo, the country’s biggest surveillance fair.

• Companies including Hikvision, search giant Baidu, and Huawei were among the 1,500 exhibitors.
• Emotion recognition is mostly being deployed at Chinese customs, where it is used to identify signs of nervousness, aggression, and the likelihood of attacking others.
• China is not the only country testing out emotion recognition capabilities, as companies including Google and Microsoft are also working on the technology.
• Nonetheless, experts say in its current state, the technology doesn’t work very well. “This technology is still a bit of a gimmick,” FT cited a popular Chinese tech blogger as saying.

Context: China’s surveillance capabilities have expanded rapidly over the past few years, with a focus on domestic stability.

• Eight of the ten most surveilled cities in the world are in China, according to a report by Comparitech, which provides resources for comparing tech services.
• The southwestern city of Chongqing came in at number one, with around 168 cameras per 1,000 people.

China’s share of global artificial intelligence (AI) investment is shrinking amid ongoing trade tensions with the US and efforts to curtail Chinese firms’ access to American components, according to new research.

Why it matters: China hopes to become a world leader in artificial intelligence by 2030, according to a 2017 plan by the State Council, the country’s cabinet.

• China overtook the US in AI-focused venture capital investment in 2017. Nonetheless, it was unseated a year later as investments in American startups more than doubled.
• Several of the country’s biggest AI firms were placed on the so-called US Entity List, effectively blocking them from doing business with American companies.

“If the Chinese AI chipset industry starts to take over, that will be China’s opportunity to reclaim the top spot. Startups like Horizon Robotics, Cambricon Technology, Unisound, and Pony.ai seem to be unicorns in the making.” 

—Lian Jye Su, principal analyst at ABI Research, told TechNode

Details: AI investment in the US totaled $9.7 billion in 2018 compared with China’s $7.4 billion, according to ABI Research. In year-on-year comparisons, US investments in AI grew 120% while those in China rose 54%.

• The US-China trade war has resulted in “collateral damage” in AI investment in China, while blacklisting Chinese AI firms has temporarily halted AI development, according to the research.
• The US accounted for 52% of global AI investments in 2018, a figure that could rise to 70% this year.
• China’s traditional industries have also been slow to adopt AI given the relatively cheap labor costs. “This means less risk-taking in AI startups that are less mature,” according to ABI.
• Meanwhile, China’s startups face challenges expanding internationally as a result of differing regulations to their home market.
• Nevertheless, the country’s favorable policies and regulations are set to drive adoption in the region, the researchers said.
• China’s venture capital market is typically responsive to new directives from the government, and updated mandates could boost investment, Su told TechNode.

Context: In early October, Sensetime, iFlytek, Megvii, and Hikvision, among others, were placed on a US trade blacklist. A number of these firms said they don’t expect the move to have a significant effect on their businesses.

• Regardless, the ban cuts off supplies of US-made hardware to these companies, whose components are used inside their products.
• At the same time, the government has scaled back spending as macroeconomic factors take their toll. According to iFlytek’s latest earnings, growth for some of its businesses suffered as a result of belt-tightening across government and private sectors.

iFlytek’s net profit for the third quarter doubled year on year to RMB 184.1 million (around $26.1 million), coming shortly after the company was added to a US trade blacklist earlier this month.

Why it matters: iFlytek was one of several Chinese artificial intelligence (AI) firms included on the so-called US Entity List, effectively blocking the company from doing business with American firms without explicit permission.

• The ban will not have a significant impact on iFlytek’s operations and development, CEO and chairperson Liu Qingfeng wrote in an internal memo to employees shortly after the ban was announced.
• iFlytek is one of China’s national “AI Champions” alongside Sensetime and Hikvision—which were blacklisted at the same time—as well as Baidu, Alibaba, Tencent, and Huawei. These companies have been tasked with spearheading China’s AI efforts as it seeks to become a technological leader by 2030.

China closes ranks as AI firms join Huawei on US blacklist

Details: While third-quarter profit increased by 108%, the company’s revenues grew by just 13% year on year, iFlytek said in a filing to the Shenzhen Stock Exchange.

• Cash received from investments made up a substantial portion of its growth, reaching nearly $1.9 billion compared with $74 million during the same period last year.
• iFlytek’s revenue for the first three quarters grew by 24% year on year to reach RMB 6.6 billion.
• The company said macroeconomic factors have impacted the growth of some of its businesses as the government, banks, and private sector have tightened their belts.

Context: iFlytek focuses on natural language processing, speech evaluation, and speech recognition. The company says it has more than 70% share of the market in China.

• iFlytek provides several products, including translation and transcription services, to Chinese consumers. It also offers its voice recognition platform to China’s healthcare and education industries, as well as to the country’s judiciary.
• The company was included on the US Entity List for its alleged involvement in Beijing’s treatment of Uighur Muslims and other predominantly Muslim ethnic minorities in China’s northeastern Xinjiang Uighur Autonomous Region.

Shanghai-based artificial intelligence (AI) startup Emotibot has closed its $45 million Series B+, aiming to improve the emotion-sensing capabilities of robots in human-machine interaction.

Why it matters: The company lists social media and gaming giant Tencent, video streaming firm iQiyi, and robot maker UBTech as its partners.

• Emotibot also works with China Merchants Bank, China Minsheng Bank, and China Mobile, according to its website.
• The company provides its computer vision and natural language processing (NLP) capabilities, as well as visual and voice emotion-recognition services to the financial, e-commerce, call center, and smart device industries.

Details: Emotibot’s latest round was led by V Fund Management, Linfeng Capital, and an undisclosed strategic investor. Puhua Capital, Keywise Capital, and VC Keywise Capital also participated.

• Emotibot said it will use the latest investment for continued NLP research and to develop its Bot Factory service, a platform that allows companies to create their own chatbots.
• Huawei has built an employee service platform using Bot Factory, which allows its staff to query issues including reimbursements and IT problems.

Context: Emotibot focuses on building chatbots that are able to identify emotional responses in humans, which the company believes is the next step in the evolution of artificial intelligence.

• Founded in 2015 by former Microsoft director Kenny Chien, Emotibot raised $30 million in its Series B last year.
• In 2017, the State Council, China’s cabinet, laid out plans to become a world leader in artificial intelligence by 2030.
• Emotibot is one of a slew of companies hoping to profit from the country’s national ambitions.

China’s biggest search engine Baidu is optimistic about its investments in artificial intelligence (AI) despite the company’s recent financial troubles and increasing competition from domestic competitors.

Why it matters: Baidu has had a rollercoaster of a year, including seeing its share price fall by 36% year to date. The company posted its first quarterly loss in the second quarter since it went public in the US in 2005.

• In January, Baidu CEO Robin Li warned that “winter is coming” referring to the troubles from a slowing economy and the fallout from a prolonged US-China trade war. The company has since implemented a number of cost-cutting measures in an attempt to mitigate risks to its business.
• Nonetheless, Baidu is pushing forward with its AI ambitions, including its costly autonomous driving program after launching a robotaxi pilot in central China.

“Artificial intelligence will not destroy human beings but will give people eternal life… Everything every person has said and done, even people’s memories, emotions and consciousness can be digitally stored on network disks or the cloud. Machines can learn people’s way of thinking.”

—Li as cited in the South China Morning Post

Details: Li was speaking at the plenary session of the World Internet Conference in the eastern Chinese city of Wuzhen on Sunday. He believes that AI will bring revolutionary changes to how humans and machines interact and act as a force for good.

• The CEO said he is optimistic about AI’s effects on society, explaining that he expects the technology to improve people’s lives.
• Li said that Baidu’s robotaxi initiative in the central Chinese city of Changsha not only serves to develop the autonomous driving industry but also forces the transport infrastructure as a whole to upgrade, the SCMP reported.

Bytedance takes on Baidu with investment in Wikipedia-like Hudong Baike

Context: In 2017, the State Council, China’s cabinet, set out goals aimed at making the country a world leader in AI by 2030. Baidu was later named one of China’s “AI Champions” for its autonomous driving efforts.

• The company has made increasing efforts to pivot toward enterprise solutions, including cloud computing and AI, as competition in the consumer market increases.
• Baidu recently invested around $200 million in Neusoft Holdings in a move that could help the company scale its AI offering.
• Tencent and Bytedance pose a serious threat to Baidu in competition for revenue as advertisers tighten their belts.
• Bytedance this year launched a search engine and invested in Wikipedia clone Hudong Baike, a competitor to Baidu’s online encyclopedia, moving further onto the search giant’s turf.

Daniel Povey, former Johns Hopkins professor and developer of open-source speech recognition toolkit Kaldi, is currently in talks to join smartphone maker Xiaomi to develop a next-generation voice recognition platform for the company.

Why it matters: Xiaomi has sharpened its focus on its smart home ecosystem, which is typically controlled using the device maker’s smart speakers and voice assistant.

•  Johns Hopkins fired Povey after he broke into a student sit-in at the university in May with bolt cutters. The students reportedly locked themselves in the institution’s administration building to protest the university hiring a private police force to maintain order on campus.
• Johns Hopkins said Povey’s actions endangered the university’s students, which led to his dismissal.

“I am very close to signing an agreement to work for Xiaomi in Beijing. Would leave before end of 2019, and would hire a small team there to work on next-gen `PyTorch-y’ Kaldi.”

—Povey on Twitter

Details: Povey, an important figure in the speech recognition field, was originally hired by Facebook after his dismissal from the university. However, he eventually declined the posting when the social media giant added conditions to his contract due to the circumstances under which he left Johns Hopkins.

• Xiaomi was not immediately available for comment when reached by TechNode on Friday.
• In a post on his personal website dated August 16, Povey said that he was seeking opportunities to work for a “Chinese company and maybe get a part-time position at a Chinese university.”
• Xiaomi’s XiaoAi voice assistant is so far only available in Chinese, though Povey’s hiring may hint at the company seeking to expand functionality to other languages.
• The voice assistant had been activated 100 million times since its launch last year, the company’s CEO Lei Jun said in January.

Context: Xiaomi adopted a strategy this year dubbed “Smartphone + AIoT,” or artificial intelligence of things, as it expanded its connected device offering, hoping to combat a slowdown in China’s smartphone market.

• Voice control is central to this strategy as the company adds voice assistant functionality to an increasing number of devices, including washing machines and smart locks.
• Xiaomi said in its first-half results that the AI assistant has around 50 million monthly active users, up 88% year on year.
• Povey is a key developer of Kaldi, an open platform used by researchers and developers for speech-related applications.

Artificial intelligence startup Megvii will continue to seek a Hong Kong listing despite being blacklisted by the US earlier this month and is aiming for an early November listing hearing, Bloomberg reported.

Why it matters: Megvii filed for a Hong Kong initial public offering (IPO) in August following reported delays due to ongoing US-China trade tensions.

• The filing came after months of political unrest in the city, with companies including e-commerce giant Alibaba opting to wait in order to gauge investor sentiment following city-wide protests.
• After Megvii’s blacklisting, Goldman Sacks, one of the IPO’s underwriters, said it was reevaluating its involvement in the company going public.
• Investors could be discouraged from buying stock in a company that has been banned from doing business with US firms.
• The government of Taiwan is opening an investigation of the firm, which was awarded in September a contract to install a security system for the Taichung Power Plant, after the blacklisting.

Details: Megvii is currently seeking a listing hearing in November, people familiar with the matter told Bloomberg.

• While being added to the US Entity List, which effectively banned Megvii from sourcing products and components from American companies, has cut off Megvii off from an essential supply of chips from Nvidia, the company has not given up on its listing goals, the people said.
• Megvii said previously that its inclusion on the blacklist came as a result of a “misunderstanding,” adding that the company complies with all laws in regions in which it operates.
• Megvii cited being put on a US trade blacklist as a possible risk in its IPO prospectus. The company said that, should it be prohibited from procuring “certain goods and technologies,” its “ability to develop and provide solutions might be impaired.”

Biometrics firm Megvii’s contract in Taiwan at risk after US blacklisting

Context: Alibaba-backed Megvii provides its facial recognition technology to companies including smartphone maker Xiaomi and payments firm Ant Financial. The AI firm also supplies solutions for public security bureaus around China.

• Megvii was one of eight Chinese AI firms blacklisted by the US for their alleged complicity in human rights violations in northwest China’s Xinjiang Uyghur Autonomous Region.
• Other companies recently included on the list include Sensetime, the world’s most valuable AI startup, speech recognition and national language processing firm iFlytek, as well as surveillance camera makers Hikvision and Dahua Technology.
• A company IPO typically takes place five to eight weeks after receiving approval from the listing committee during the hearing, according to guidelines from corporate law firm Mayer Brown. Xiaomi passed its listing hearing on June 7, 2018 and floated its shares on the Hong Kong stock exchange on July 8, 2018.

The Taiwan government announced that it would investigate Chinese artificial intelligence (AI) and biometrics company Megvii after it was blacklisted by the US government last week, Focus Taiwan reported. 

Why it matters: The investigation could put at risk Megvii’s recent contract to install a biometric security system at Taiwan’s Taichung Power Plant, with legislators wondering whether the firm’s technology raises national security concerns. 

• It could also complicate Megvii’s plans for its initial public offering (IPO), which, according to co-founder and CEO Yin Qi, it planned to “firmly carry on” with despite “some disturbance” after being blacklisted along with 27 other Chinese government agencies and companies.
• Yin also said the company has been preparing for possible supply shortages since May in anticipation of being blacklisted and barred from purchasing crucial x86 servers and GPUs subject to US export regulations. 

Details: In response to the Economic Committee’s concerns, Minister Sheng Jong-chin said he would ask the Ministry of Economic Affairs (MOEA)’s State-Owned Enterprise Commission to launch a comprehensive investigation of the blacklisted companies and instruct the ministry’s Information Management Center to conduct a review.  

• Megvii won a bid to install the security system at Taichung Power Plant on Sept. 12, less than a month before being added to the US government’s blacklist. 
• The system uses facial recognition to monitor contractors entering and leaving the facility. 
• According to Shen Jong-chin, it does not include employees of the state-run Taiwan Power Company.  

Context: Megvii joins seven other Chinese companies and 20 of the country’s government agencies on the blacklist, including the police bureau in the Xinjiang Uighur Autonomous Region.

• Beijing has since demanded that the US reverse its decision but hasn’t said whether it will retaliate. 
• Goldman Sachs said in a statement Tuesday it was reviewing its involvement as a joint sponsor of Megvii’s IPO alongside JPMorgan Chase and Citigroup, according to CNBC. 
• The Alibaba-backed AI firm is looking to raise up to $1 billion in a listing on the Stock Exchange of Hong Kong.  

The US move to blacklist Chinese artificial intelligence and surveillance companies has met with widespread condemnation in China despite the companies’ claims that the prohibition won’t have long-term effects on their operations.

On October 7, the US Commerce Department placed several high-profile Chinese AI firms and government agencies on the Entity List—effectively blocking them from doing business with American companies without prior approval.

The organizations include Sensetime, the world’s most valuable AI startup, and the speech recognition and natural language processing firm iFlytek, as well as surveillance camera makers Hikivision and Dahua Technology.

Also included are Alibaba-backed Megvii, which recently filed to go public in Hong Kong—the first Chinese AI startup to do so—and its counterpart Yitu.

In its statement, the Commerce Department said that the companies have been “implicated in human rights violations and abuses” in northwest China’s Xinjiang Uyghur Autonomous Region, home to predominantly Muslim ethnic minorities. According to US officials, the blacklisted companies were involved in creating the mass surveillance apparatus used to monitor minorities in the region.

Sensetime, Megvii, Yitu, and Hikvision all issued statements opposing the US decision, saying that they comply with Chinese laws. Megvii claimed its inclusion on the list was the result of a “misunderstanding.”

The US announcement came days before Chinese vice-premier Liu He’s visit to Washington to resume trade talks. The Commerce Department maintains that the ban is unrelated to the negotiations.

The blacklisting of these companies is a direct affront to China’s technological ambitions. The country has plowed billions of yuan into shifting its economy up the industrial value chain through the development of the robotics, electric vehicle, semiconductor, and AI sectors. Given that China’s goal is to overtake the US by becoming an AI frontrunner by 2030, the Commerce Department’s move appears to takes aim at that ambition.

While US human rights and pro-democracy groups applauded the export restrictions, the conversation within China has remained focused on the affected companies. Meanwhile, nationalistic fervor has reached fever pitch, with state media claiming the move is just the latest attempt by the US to undermine China’s rise.

Support at home

China swiftly hit back at the ban, saying that the human rights claims are “fact-distorting gibberish” that attempt to limit China’s counterterrorism efforts.

“Xinjiang does not have the so-called human rights issue claimed by the US. The accusations by the US side are merely made-up pretexts for its interference,” said Geng Shuang, China’s Foreign Ministry spokesperson, on Tuesday.

Meanwhile, state-backed Xinhua News Agency said the ban was aimed at “hindering China’s development,” rather than any real concerns over human rights.

The Global Times, a Communist Party mouthpiece, referred to the ban itself—as well as subsequent visa restrictions on Chinese officials believed to be involved in human rights violations in Xinjiang—as “shameful,” adding that the sanctions will have “little effect.”

Much of this rhetoric was mimicked on Chinese social media, where the majority of netizens put their support behind the homegrown companies.

“America is worried about the rise of great Chinese companies,” one user said, commenting on a statement published by Sensetime on the popular messaging app WeChat.

“Without the US Entity List, many people wouldn’t know that Chinese innovation is leading the way,” said another.

On Weibo, users attributed the ban to the perceived fear of China in the US.

“When the US wildly starts boycotting China, it shows China is powerful enough to scare them,” commented one user from eastern China’s Shandong province.

Meanwhile, the Washington-based NGO Freedom House saw the move as a long-awaited offensive against alleged human rights violations in Xinjiang.

The organization said in a statement on Wednesday that it “applauds” the decision, urging the US government to expand the export restrictions to encompass more companies involved in surveilling dissidents in China.

The Republican senator Ted Cruz, who—as part of a bipartisan coalition of US senators—has pushed for sanctions against entities suspected of involvement in human rights violations in Xinjiang, called the move an “excellent step forward.”

What next?

Despite the polarizing effect of the ban, the prohibition did not come as a surprise. Megvii, which provides its facial recognition technology to smartphone maker Xiaomi and payments firm Ant Financial, had alluded to the risks of a potential blacklisting when filing for its Hong Kong IPO.

“If we were subject to economic and trade restrictions, we could be prevented from procuring certain goods and technologies, and our ability to develop and provide our solutions might be impaired.” the company said in its prospectus.

Megvii cited Huawei as an example. Earlier this year, the Chinese telecommunications giant was added to the Entity List over alleged national security concerns related to the company’s technology.

Other companies, including iFlytek and Hikvision, began making contingency plans as far back as last year.

Hikvision said it had begun seeking alternative suppliers to limit its reliance on the US in late 2018, the same year that Chinese telecommunications giant ZTE was banned from sourcing American components for violating US sanctions against Iran and North Korea. ZTE has subsequently been taken off the blacklist.

In an internal memo to employees on Wednesday, iFlytek CEO Liu Qingfeng said the company had already made plans to deal with the situation, adding that he still expected positive financial results for the remainder of the year.

The overall impact on companies such as Sensetime, Megvii, and Yitu is likely to be limited. Megvii’s overseas business accounted for less than 5% of its total revenue during the first half of 2019. Sensetime and Yitu’s financial records are not currently available, as the companies have not submitted listing documents.

Nonetheless, being added to the blacklist could have severe reputational impacts. Chinese AI firms have established partnerships with universities around the world, which these academic institutions are now rethinking.

In June, prior to the export ban, the Massachusetts Institute of Technology (MIT) reportedly began reassessing its partnership with iFlytek as a result of the speech recognition firm’s technology being used in Xinjiang. The review came after Rutgers University severed its relationship with the company.

Just last week, MIT began reassessing its partnership with Sensetime after the firm was added to the blacklist. Sensetime has similar partnerships with research institutions as well as research centers and offices in Japan, Singapore, and the United Arab Emirates.

The effect on companies like Hikvision, which has become the largest supplier of surveillance equipment in the world, could be more pronounced. Around 30% of the company’s operating income currently comes from its overseas sales, according to figures for the first half of the year.

In addition, the company’s components are sourced from a vast array of US manufacturers, including Intel, Seagate, Nvidia, and Western Digital, according to John Honovich, the founder of IPVM, a video surveillance research company. But the company also has a number of non-US suppliers.

Nevertheless, Hikvision executives claim that the company’s reliance on US technology is “relatively low,” instead opting to use semiconductors from companies including Huawei’s chipmaking subsidiary HiSilicon, among others.

Speech recognition firm iFlytek’s operations and development will not be significantly affected by the company being included on a US trade blacklist, according to its chief executive officer.

Why it matters: iFlytek along with several other Chinese technology companies and government agencies were on Monday added to the so-called US Entity List, effectively banning them from doing business with American firms.

• The US Commerce Department said the blacklisting comes as the companies are “implicated in human rights violations and abuses” in China’s western Xinjiang Uigur Autonomous Region, home to predominantly Muslim minorities.
• Sensetime, the world’s most valuable AI startup, and surveillance camera maker Hikvision, as well as Megvii and Yitu, were also added to the list.
• iFlytek is one of China’s national “AI Champions” alongside Sensetime, Baidu, Alibaba, Huawei, and Hikvision, among others. These companies are tasked with spearheading China’s AI efforts as it seeks to become a technological leader by 2030.

Details: iFlytek chairperson and CEO Liu Qingfeng wrote in an internal memo on Wednesday that the company would appeal its inclusion to the Entity List, adding that the iFlytek would see healthy growth throughout the rest of the year.

• The company expects to report profits of between RMB 330 million and RMB 380 million in the first three quarters of 2019, up 50% to 73% year on year, according to documents filed to the Shenzhen Stock Exchange on Thursday.
• Liu said that iFlytek has made it possible for people to “communicate freely” using its translation services, with its products being used in 200 countries and regions around the world.

Artificial intelligence (AI) unicorn Sensetime is partnering with communications tower company China Tower to retrofit its 1.9 million sites with video surveillance systems.

Why it matters: At a recent trade show, China Tower offered a range of scenarios where the monitoring tech could be useful, including forest fire prevention and curbing illegal farmland construction.

• A large-scale, AI-powered surveillance network could raise concerns from privacy advocates worried about China’s already massive security apparatus.
• The partnership provides a boost for Alibaba-backed Sensetime, which is China’s biggest AI startup with a valuation exceeding $7.5 billion.

Details: The new system is part of China Tower’s Trans-sector Site Application and Information business (TSSAI).

• A new subsidiary named Smart Tower that was set up in June will drive the surveillance business, according to a report by communications industry media Light Reading. Smart Tower has registered capital of $130 million.
• China Tower has also contracted Shanghai-based China Pacific Property Insurance to analyze insurance claims using AI.

Context: China Tower previously partnered with Alibaba in a similar infrastructure-oriented deal, lending its towers in return for collaboration on cloud computing, edge computing, big data, and 5G.

• Along similar lines, in 2018 China Tower joined the Ministry of Industry and Information Technology’s (MIIT) pilot program to explore potential uses for retired electric vehicle batteries, technology it already leverages to provide backup power for its base stations.
• Tower firms outside China have been slow to adopt China Tower’s business strategy, with the exception of Barcelona-based Cellnex, which offers its infrastructure for services like Wi-Fi, irrigation management, and the internet of things.

E-commerce giant Alibaba has released a self-developed artificial intelligence (AI) chip, as the company increases its focus on chipmaking and aims to improve efficiency on its shopping platforms.

Why it matters: The chip, a neural processing unit, is developed by Alibaba’s chipmaking subsidiary T-Head, known as Pingtouge in Chinese. The company was set up in September last year.

• China has set its sights on driving its domestic semiconductor industry while simultaneously pushing to become a leader in AI by 2030.
• Observers have said that China’s lack of hardware prowess is currently preventing the company from overtaking the US in the race for AI supremacy.

“The launch of Hanguang 800 is an important step in our pursuit of next-generation technologies, boosting computing capabilities that will drive both our current and emerging businesses while improving energy efficiency.

—Jeff Zhang, Alibaba Group chief technology officer

China’s ‘military-civil’ partnerships could hurt its AI ambitions: report

Details: While Alibaba is currently using the Hanguang within its own operations, the company plans to make the chip’s computing power available through its cloud services.

•  The chip “largely outpaces” the industry average, Alibaba said in a statement.
• The company added that the task of classifying and tailoring product recommendations for the 1 billion images uploaded every day to e-commerce platform Taobao used to take one hour, with the new chip cutting that time to five minutes.
• The chip is currently being used to optimize product search and automatic translations on Alibaba’s e-commerce sites, and is also being applied to personalized recommendations and advertising, the company said.

Context: T-Head was formed under Alibaba’s research and development unit DAMO Academy in late 2018. The company earlier this year released an internet of things processor based on RISC-V, the open-source instruction set architecture.

• Alibaba has made investments in s slew of chipmakers, including China-based Cambricon, Kneron, ASR, and DeePhi.

China’s internet regulator has instructed online platform operators ensure that their content recommendation algorithms create a “healthy” and “positive” online environment, according to draft rules released on Tuesday.

Why it matters: The Chinese government has taken an increasingly heavy hand when dealing with online content. Beijing has accelerated efforts to rid the internet of “inappropriate content.” Few of the country’s tech companies have managed to avoid censure during the campaign.

“Online platforms should strengthen management of information recommendation or presentation by methods including manual editing or machine algorithms to create a positive and healthy ecosystem.”

–Cyberspace Administration of China (CAC)

Details: The CAC’s draft regulations cover a wide range of online platforms from websites to apps and online forums. The document is open for public comment until October 10.

• Recommendations by algorithms should not include content that undermines national unity, disseminates false information, subverts the national regime, or disrupts economic order.
• However, the CAC promotes content that publicizes Xi Jinping Thought, socialist core values, and increases the influence of Chinese culture on the international stage.
• Platforms are required to create systems to conduct real-time inspections, respond to emergencies, and handle online rumors, with the rules consolidating previously implemented measures.
• Companies will also need to improve regulation of “hot topics” and top search terms to ensure they don’t contain prohibited content.
•  Users should take an active role in governing online platforms through complaints and reporting illegal content, the CAC said.

Context: Operators of services ranging from dating apps to short video platforms have all been censured for hosting “vulgar” content. Companies including Tencent, Weibo, Baidu, and Bytedance have all been affected by an extended operation to clean up China’s cyberspace.

• Regulators have also sought to expand social credit blacklists to online platforms and their users to punish conduct that is deemed to be untrustworthy.

The Trump administration’s chief technology officer warned on Tuesday that while the US currently leads in artificial intelligence (AI) development, China is quickly narrowing the technology gap.

Why it matters: China’s government has taken a top-down approach to improving the country’s technological capabilities, with emphasis on AI advancements.

• The State Council, China’s cabinet, in 2017 laid out plans to become a global leader in AI development by 2030.
• Some observers have dubbed the US-China dynamic an “AI arms race” amid trade tensions between the world’s two largest economies.

“Although America is the leader in AI,  China is working to catch up… Today, our goal is very clear: The uniquely American ecosystem must do everything its collective power can to keep America’s lead in the AI race and build on our successes.”

—US Chief Technology Officer Michael Kratsios

Details: Kratsios added that the competition between the two nations too often focuses on the disparity in government spending on research and development, referring to China’s funding budgets as being “aspirational” and “cryptic.”

• Kratsios was speaking at an event in organized by think tank the Center for Data Innovation (CDI) in Washington D.C.
• He said that the US is home to 17 of the world’s 32 AI unicorns, adding that the country has around 2,000 AI companies. The CTO criticized the Chinese government for “choosing winners in the AI field,” referencing the country’s state-sanctioned AI champions—Tencent, Baidu, Alibaba, Sensetime, and iFlytek.
• Kratsios said US government agencies planned to spend $1 billion on non-defense AI research during the next fiscal year.

Context: In February US President Donald Trump signed an executive order directing government agencies to increase their focus on AI. However, observers criticized the order, saying it lacked clarity and funding goals.

• Kratsios is not the first to warn of China’s technological rise. In a report last month, the CDI said that China lags in AI but is catching up to the US.
• The organization said the country’s civil-military partnerships, in which the government is promoting closer ties between the private sector and the military, could hurt China’s AI ambitions as distrust of companies linked to the Chinese government grows amid US-China tensions.

Researchers at Chinese gaming giant NetEase published a paper detailing a new machine learning method that enables players to create in-game characters from a selfie, Synced reports. 

Why it matters: The researchers frame the value of their work as a means to streamline the often-laborious character customization process in contemporary RPGs. 

• As game developers look to set their games apart through the level of personalization they offer to their players, the added immersion that comes with translating one’s face into an avatar could also prove valuable. 

Details: Titled “Face-to-Parameter Translation for Game Character Auto-Creation” and published September 3 on a publishing platform associated with Cornell University, the paper describes how a deep generative network can transform a portrait into a character whose style matches that of the desired game engine. 

• To improve accuracy and enable further customization, the 3D face reconstruction method creates a bone-driven model, as opposed to previous methods that created 3D face mesh grids. 
• In addition to photographs, the generator also works with sketches. 

Context: Following the recent backlash against deepfake app Zao for its excessive data collection, it is unclear how wider audiences will respond to a big tech company soliciting this type of personal information. 

• The technology has already been used over one million times by Chinese gamers. 
• NetEase isn’t the first in the entertainment industry to explore the potential of artificial intelligence—in August, researchers at Baidu’s iQiyi created a facial recognition dataset based on anime characters. 

Search giant Baidu will invest RMB 1.4 billion (around $200 million) in technology investment firm Neusoft Holdings, as the pair look to develop smart city, healthcare, and education solutions.

Why it matters: Baidu is one of the biggest artificial intelligence (AI) companies in China and could see a partnership with Neusoft as a means to better scale its products.

• Baidu has shown continued interest in Neusoft over the course of the year, signing a partnership deal with Neusoft Corporation and investing in Neusoft Medical Systems, subsidiaries of Neusoft Holdings.
• The internet giant is looking for ways to make up for slowing ad revenue as advertisers tighten their belts amid a slowing economy and regulators crack down on online content.

Details: The deal will make Baidu chief technology officer Wang Haifeng a board director at Neusoft.

• In the medical field, the companies will work together to develop services including AI medical diagnosis, hospital cloud services, medical big data, and healthcare management, our sister site TechNode Chinese reported.
• Meanwhile, in education, Baidu and Neusoft plan to develop AI-driven education services as well as online courses.
• The companies will also work on smart city, smart transport, smart government, and industry cloud applications.

Context: Baidu’s year got off to a difficult start after reporting a quarterly loss for the first time since listing in 2005.

• The company beat revenue expectations in the second quarter, but its net income fell by 60% year on year.
• Bytedance and Tencent pose a serious threat to Baidu, as both companies have short video apps that have proven to be a major draw for advertisers.
• Bytedance moved further onto Baidu’s turf after launching a search engine last month and investing in Wikipedia-like Hudong Baike, competitor to Baidu’s own online encyclopedia.

Artificial intelligence (AI) startup Yitu is considering a listing on China’s new Nasdaq-style tech board, Bloomberg reported, citing people familiar with the matter.

Why it matters: News of the possible listing comes shortly after rival AI firm Megvii submitted an IPO prospectus to the Hong Kong stock exchange, the first Chinese AI company to make such a move.

• AI technologies from both companies are used widely in China and they are key suppliers to the country’s public security organs.
• China is home to a growing number of AI startups, several of which have their eyes on capital markets.

Details: Yitu reportedly plans to list on Shanghai’s Star Market as early as this year, according to the Bloomberg report.

• No details of the listing have so far been revealed. The sources did not indicate how much Yitu is hoping to raise by going public.
• Since Yitu has not officially filed for an IPO, its financial position is unclear. However, Star Market allows unprofitable companies to apply for a listing.

Alibaba-backed AI startup Megvii files for Hong Kong IPO

Context: The IPO plans come amid government calls to stimulate China’s high-tech development, with ambitions to become a world leader in AI by 2030.

• The country is home to the world’s most valuable AI startup, Sensetime, though the company has not made public any plans for a listing.
• Before filing, Megvii had reportedly been reconsidering its IPO plans as a result of the ongoing US-China trade tensions.

Bottom line: China’s AI companies are looking to tap capital markets as venture funding slows and investors raise questions about the prospective profitability of these firms.

A study by Hong Kong-based Insilico Medicine utilized a new AI system that created a series of novel molecules capable of treating fibrosis and other illnesses on a significantly shorter timeline, the South China Morning Post reported. 

Why it matters: Insilico used AI to identify treatment candidates in just three weeks compared with the traditional drug discovery process, which can take decades and cost billions. Besides saving money for pharmaceutical companies, streamlining the development of new drugs could also save lives. 

Details: Out of the six molecules that the firm successfully created, one was later found to be effective in treating mice with renal fibrosis. 

• Insilico’s system dubbed GENTRL is powered by a “generative chemistry that utilizes modern AI techniques” similar to Google’s Go-playing DeepMind, according to BioSpace.
• GENTRL was developed in collaboration with pharmaceutical services contractor Wuxi Apptec and the University of Toronto’s Alán Aspuru-Guzik. 
• The pharmaceutical research firm’s paper describing its success was published in the journal, Nature Biotechnology, on Monday. 

Context: With China’s healthcare data industry set to top RMB 80 billion by 2020, artificial intelligence applications show promise in the face of a rapidly aging population coupled with a dearth of doctors to provide care. 

• Tencent Holdings has partnered with more than 100 Chinese hospitals to research new AI applications.
• Alibaba is also collaborating with hospitals to develop smart diagnosis platforms.  
• Baidu is developing an open-source AI platform to assist in breast cancer detection. 

Despite calls to regulate artificial intelligence on the battlefield, China’s tech sector is in danger of complicity in developing lethal autonomous weapons, as several companies have shown a keen interest in collaborations with the country’s public security organs.

Sensetime, the world’s most valuable AI startup, and facial recognition firm Yitu were cited by Dutch anti-war non-governmental organization PAX over concerns that their technology could be used for developing “killer robots” that could choose and engage targets without human intervention.

While Sensetime and Yitu’s products are currently not employed on the battlefield, the nature of those products as well as the companies’ history of working with China’s government is worrying, PAX says. In a recent report, the NGO referred to the two firms as being of “high concern.”

Sensetime and Yitu were not immediately available for comment.

PAX’s report ranks the possible complicity of tech companies according to the technologies they develop, past collaboration with law enforcement or the military, and whether they have pledged not to aid in the development of killer robots.

The report also mentions other Chinese tech firms, including Alibaba, Baidu, and Tencent, though PAX classifies these companies as less of a concern.

PAX’s report comes amid increasing calls for caution over what has been dubbed the third revolution in warfare, after gunpowder and nuclear weapons. Around 30 countries currently support a ban on killer robots, and prominent figures from the research and tech communities, including Tesla’s Elon Musk, who spoke at a government-led AI conference in Shanghai this week, have warned of the dangers they present.

Currently, seven nations are developing lethal autonomous weapons, including the US and China. The projects under development include autonomous drones, as well as AI-equipped tanks and fighter jets, whose autonomy have raised alarm bells.

“Killer robots would be unable to apply either compassion or nuanced legal and ethical judgment to decisions to use lethal force,” Human Rights Watch said of the technology earlier this month.

In the US, tech companies including Google and Palantir have taken on government contracts, with applications ranging from analyzing drone footage to documenting immigrants. The same is true in China, where the private sector has filled government tenders to provide technology in a bid to ensure social stability.

PAX’s report raises questions over possible tech sector involvement in the race for the next generation of military technology, in which lucrative government contracts could provide significant incentives. Meanwhile, China holds an ambiguous stance toward autonomous weapons, supporting a ban on these arms while simultaneously pushing for the prohibition to exclude developing such weapons.

“It’s very clear that the Chinese military is very actively engaged in pursuing a number of applications of AI,” says Elsa Kania, an adjunct senior fellow who studies the modernization of China’s military at Center for a New American Security, a Washington DC-based think tank.

The future of combat

“In future battlegrounds, there will be no people fighting,” said Zeng Yi, a senior executive at Norinco, one of China’s biggest defense companies, at the Xiangshan Forum in Beijing last year.

The Xiangshan Forum is a big deal. With its focus on security in the Asia-Pacific region, it is to the Shangri-La Dialogue what the Boao Forum is to Davos. And Norinco is a key player in China’s defense industry; its products are used both domestically and internationally, including in the Middle East.

Zeng went on to predict that by 2025, autonomous weapons would be ubiquitous on the world’s battlegrounds, given the use of AI. “We are sure about the direction and that this is the future,” he added.

This kind of thinking has critics concerned. Much like the sprint to produce nuclear weapons during the Cold War, a push to develop autonomous weapons could lead to what PAX calls an “AI arms race,” in which various states compete to develop these weapons. Unlike nuclear arms, which act as a deterrent, autonomous weapons could make nations increasingly trigger-happy, as “you don’t have to put troops on the groups,” observers say.

Like most sectors earmarked for development, the government has put its might behind modernizing the military, creating an attractive proposition for tech startups. Daan Kayser, PAX’s project leader on autonomous weapons, told TechNode in a phone interview, “For Chinese companies, these could be quite lucrative projects, so there are economic reasons for getting involved.”

Financial incentives are evident in China’s surveillance sector, where companies like Sensetime, Yitu, and rival Megvii—which this week announced plans for a Hong Kong listing—have seen their profits swell on the back of government contracts.

While financial figures aren’t available for Sensetime and Yitu, documents filed with the Hong Kong Stock Exchange show that Megvii’s revenue reached almost RMB 1 billion ($133 million) in the first half of 2019, which the company attributes, in part, to government spending. The AI firm’s revenue in the first six months of this year was three times that of sales for the whole of 2017.

Sensetime, Yitu, Megvii, and Cloudwalk—also mentioned in PAX’s report—have all developed AI monitoring systems that help China’s police force keep tabs on its citizens by analyzing video and flagging persons of interest.

For example, Sensetime’s SenseTotem and SenseFace systems are currently being used by various police departments around China for this purpose. Meanwhile, Yitu’s tech is being used by public security organs in 20 provinces throughout the country.

“The government creates lucrative business opportunities by including these companies in its digital agenda. The companies, in turn, help secure political stability,” Sebastian Heilmann, the founding president of the Mercator Institute for China Studies, wrote in a blog post.

China’s government has also launched several state-driven investment initiatives focusing on private sector-military partnerships. As of the middle of this year, these funds had reached tens of billions of yuan.

Incentives to provide tech for killer robots could extend beyond monetary gain, as the Chinese government aims to promote an atmosphere of “civil-military fusion.” China’s army is looking to develop closer ties with the country’s private sector and research institutions.

China sees a need for these partnerships to drive a defense industry that has traditionally been viewed as unimaginative and a military that hasn’t been able to leverage commercial sector innovation. The enterprise is being overseen at the highest level, with Chinese president Xi Jinping leading the charge.

“Whenever there is a national initiative, there is pressure on companies to engage,” Kania said. She added that the military’s drive to forge close ties with civil society creates “more programs, and avenues, and opportunities” for businesses to work with the armed forces.

Despite rising pressure, companies are not being coerced into these sorts of partnerships. The characterization that the Chinese military has direct access to technology in the commercial sector is not accurate. Some tech companies have articulated interest in this type of work, while others have not—at least based on public information, Kania says.

Nevertheless, there is “absolutely a connection” between security and defense applications, she added, meaning that it wouldn’t be a stretch for companies that are involved in one to explore the other.

War games

AI is central to developing autonomous weapons, and China is betting big on the technology. The country is catching up with the US and has overtaken the European Union in its capabilities, according to the Center for Data Innovation, a US-based think tank. The State Council, China’s cabinet, has announced plans to become a world leader in AI by 2030.

Meanwhile, the country’s Made in China 2025 initiative, which the country’s leaders have touted as the strategy for moving China up the industrial value chain, prioritizes the development of the robotics, aerospace and information technology industries. All of these sectors develop dual-use, military-civil technologies.

“The Chinese military believes that there is a revolution in military affairs underway in which AI could be critical to future military power,” Kania said.

These systems could be used in applications ranging from cyberdefense to creating weapons with ever-increasing levels of autonomy. Machine recognition, in particular, could prove to be extremely valuable in developing highly autonomous weapons, allowing these arms to not only “see” the world around them, but also to understand it and make decisions based on what they perceive.

The danger, according to Kayser, is that intelligent weapons could make decisions at a speed that is out of the realm of human capability. “If you can make decisions faster than your enemy, you’ll be able to beat them,” he said.

Opposition

In June 2018, US search giant Google announced it would not renew a Pentagon contract to analyze drone video footage. Dubbed “Project Maven,” its aim had been simple: to use machine learning to improve the accuracy of drone strikes.

The tie-up came to an abrupt end. Thousands of Google’s employees signed a petition imploring the company to abandon the project, while many others resigned.

“We believe that Google should not be in the business of war,” the open letter to Google CEO Sundar Pichai began.

Similarly, the data analysis firm Palantir, founded by Facebook board member Peter Thiel, recently found itself at the center of controversy for its contracts with the US Immigration and Customs Enforcement to gather information about undocumented immigrants.

Similar opposition to tech companies working with the government has largely been absent in China.

“To my knowledge, there has not been any Chinese tech company that has been working with the Ministry of Public Security or the military where there has been any articulation of resistance to that engagement,” Kania said.

Regardless, countries working on these sorts of weapons are unlikely to stop as a result of public outcry. If even one nation pursues autonomous weapons, others will likely follow suit.

“These countries are looking at each other. The main rationale for exploring these sorts of technologies seems to be: ‘Our adversaries are also doing this,’” said Kayser.

This year’s theme for the World Artificial Intelligence Conference (WAIC), one of the largest tech events organized by the Chinese government, was “Intelligence Connectivity Infinite Possibilities.” Held in Shanghai, it provided a stage for the world’s tech giants to showcase the newest innovations. This year there was a clear focus on industrial internet applications in combination with AI technologies.

The Smart Cafe at the Amazon Web Services (AWS) booth gave a demonstration on how AI technologies can ensure that every step in a supply chain is functioning.

Didi displayed one of its robotaxis at the exhibition. Zhang Bo, CEO of Didi’s autonomous driving company, announced Friday that it will launch a pilot robotaxi fleet in Shanghai. The company declined to provide a specific timeline.

Tencent’s Intelligent Inspection and Manipulation Robot can perform dexterous movements in a special environments such as in the oil or gas industry without the need for custom code.

Crew Interactive MObile CompanioN (CIMON), the first AI assistant for humans in space, made its first appearance at WAIC this year. CIMON was designed by a collaboration between IBM, German Aerospace Center, and Airbus.

In cooperation with Xiaoguan Tea, IBM’s Visual Inspection technology can scan and pick out the impurities from tea leaves.

The vast majority of people underestimate the capabilities of artificial intelligence (AI) and its possible effects on the future of humanity, Tesla CEO Elon Musk said at a government-led conference in Shanghai on Thursday.

Why it matters: The Tesla and SpaceX CEO was debating Alibaba founder Jack Ma on the merits of AI during the opening ceremony of the World Artificial Intelligence Conference (WAIC).

• Musk has long warned of the dangers of AI and the possibility of an intelligence explosion, in which computers quickly develop intellect that far exceeds human capability.
• Governments around the world are betting on the future of the technology in industries ranging from personal finance to public security.

“There’s just a smaller and smaller corner of intellectual pursuits that humans are better than computers and every year it gets smaller. Soon we will be far surpassed in every single way.”

—Elon Musk at WAIC

Details: While Musk took his usual pessimistic approach to issues of AI, Ma spun a more optimistic thread, saying he’s not worried about the evolution of the technology.

• Ma said that he has confidence in artificial intelligence because it could help humans to understand themselves better. What solutions to issues brought about by artificial intelligence we don’t have today, we will have tomorrow, he said.
• The Alibaba founder said that the rise of AI would lead to a 12-hour work week, where people focus on jobs that they enjoy. Ma earlier this year voiced his support for 996 culture—working from 9 a.m. to 9 p.m., six days a week—saying that the hours are a “blessing.”
• Musk advised that individuals should focus on engineering or physics, adding that the last job will be that of an AI software engineer.
• He likened the coming relationship between people and AI to that of chimpanzees and humans, where people would be incapable of comprehending the actions of an intelligence smarter than themselves.
• Both founders agreed that education is imperative, with Ma saying that the current education system is stuck in the industrial era.

Context: The future of human-machine interaction has garnered much attention over the past few years. Concerns over algorithmic bias, unemployment, and the future economic system have come to the fore, with experts calling for these issues to be dealt with before its too late.

• Beijing recently released a set of principles to guide the development of AI. The guidelines detailed ways to approach the research and development, implementation, and governance of AI.
• China currently lags behind the US in terms of its AI capabilities. The country currently lacks talent and hardware prowess to propel it into the top spot. However, China sets itself apart in AI deployment and its wealth of data.

DeepMind’s losses of $570 million last year laid bare the challenges facing the artificial intelligence sector. But the Alphabet-owned company is not unique in the land of AI industrialists. Baidu, the most ardent AI advocate in China, may have lost a lot more. Worse yet, its wager on AI may go the way of its failed online-to-offline push from several years ago.  

There is no question that companies must invest in AI to succeed in the future, and losses are natural in the initial stages of developing and commercializing such an emerging technology. For Alphabet, which pulled in $137 billion last year with operating income of $26 billion, sustaining over half a billion dollars in losses at DeepMind is a reasonable price to remain competitive.   

But for Baidu, with a stock price that has fallen by more than half over the past five years (Alphabet’s share price doubled in the same period), its margin for error in AI is significantly narrower. The failure or success of its AI strategy will play a more critical role in determining the future of Baidu.

The key questions are: How much has Baidu burned in its AI bet? What do potential returns on such an investment look like? NASDAQ-listed Baidu does not break down the financial performance of its AI operations in its filings. But based on estimates from public disclosures and industry sources, Baidu could be losing billions of dollars on its AI push. Worse yet, the possibility of a handsome return appears dim.  

Road to nowhere

Let’s first look at Baidu’s investments in AI. The company has directed significant funding into developing two AI projects—autonomous driving and smart speakers. 

Baidu’s Apollo autonomous driving unit is its most capital-intensive. The firm runs the largest fleet of autonomous driving test vehicles in China. According to a report from the Beijing government, Baidu has 45 registered test vehicles in the capital. The total number of test vehicles across China, registered and unregistered, could be several hundred, according to sources. The cost of setting up an autonomous vehicle is around RMB 1 to 2 million (about $140,000-$280,000). Using a conservative estimate of 200 test vehicles at mid-range per-unit cost, that amounts to Baidu’s having spent RMB 300 million on just setting up its test fleet.

Baidu’s Apolong autonomous buses, which aim to provide shuttle services in closed environments like parks at around 10km per hour, cost over RMB 2 million per unit, a figure I have confirmed with my sources. With a list price of RMB 1.5 million, Baidu loses RMB 500,000 for each bus sold. Last year, the company said it achieved “volume production” as its 100th bus rolled off the production line. That means RMB 200 million in investments, and potentially RMB 50 million in losses if Baidu sells all of those vehicles. The actual loss could be higher, as some buses are still sitting in warehouses and each of them requires maintenance and after-sales services, according to sources.

The company has also secured around 100 autonomous driving pilot licenses in China, five times that of the next industry player, according to its Q2 earnings report. Considering that obtaining such permits requires a lot more capital in China than elsewhere due to mandatory testing at expensive designated centers, this represents another major expenditure.

Then, there is the cost of expensive AI talent: between 1,500 and 2,000 for the Apollo team. As the average annual salary for AI engineers in China is around RMB 360,000, according to jobs listing company Boss Zhipin, the salary cost of the team is at least RMB 630 million a year. Actual costs could be much higher as Baidu tends to hire top-tier talent and runs a large US team, whose salaries are much higher.    

It’s impossible to estimate Baidu’s total AI losses: It’s a complex accounting effort that can be only done by its internal finance team, dependent on thousands of accounting decisions. What I listed, including both AV and speaker amounts to roughly $500m, and that’s just a very small part. 

Perhaps the most direct yardsticks for Baidu’s total losses in autonomous driving are its peers. GM Cruise is the only notable autonomous driving unit that has disclosed its financials. It suffered total losses of $1.5 billion for the three years 2016-2018. Baidu Apollo has around 1,900 staff (this number has dropped as Baidu is reportedly shrinking the team size), more than that of GM Cruise, which is over 1,000. Baidu’s testing fleet is also bigger than that of GM Cruise (less than 200 vehicles). It would be plausible to ascertain that Baidu is suffering greater losses than GM Cruise’s $1.5 billion for the same period.

Baidu has made two critical mistakes in self-driving: First, it should have spun off its self-driving unit, like Alphabet did to form Waymo in 2016 or like Didi Chuxing did two weeks ago. The industry now realizes that it will take much longer than initially anticipated to realize L4 autonomy and the robotaxi business model. Apollo once aimed to achieve full autonomy on highways and urban roads in 2021. Nobody in the industry believes that target is remotely feasible. If Baidu had spun off Apollo last year, it could have raised capital at very attractive valuations. The longer it waits, the more realistic investors will become when considering prospects for self-driving and its price tag.

Second, Baidu’s bet on the Apollo Open Platform is an idea that is going nowhere. Taking a page from Google’s Android, Baidu created the Apollo Open Platform in 2017, hoping to become the Android for self-driving. Two years later, it’s clear that nobody—from traditional carmakers and autonomous driving startups to tech providers—wants to share their data with others. It’s also questionable whether sharing data across different locations, vehicle models, and road conditions makes sense for advancing self-driving. One source said that Baidu Apollo is an open ecosystem with only one player: Baidu Apollo itself. Most of the 145 Apollo partners are participating only nominally. 

Full of sound and fury

Baidu’s second huge AI expenditure is smart speakers. According to Canalys, Baidu shipped 11.6 million smart speakers from 2018 to the second quarter this year. As Baidu prices its smart speakers below cost to grab market share, the company is losing around RMB 200 per unit sold. That equates to RMB 2.3 billion in total losses just on speaker sales alone.

Compared to its competitors, Baidu has little chance of turning around these losses: it appears to be fighting for an entry point leading not to its strengths, but to its weaknesses. Baidu’s vice president at its smart living group Jing Kun said in an interview that Baidu will first try to monetize smart speakers by providing content-based subscriptions. But content is not Baidu’s strength in light of Toutiao’s incursion onto its turf. Apple, for example, aims to profit from its speakers by bolstering Apple Music subscriptions. Tencent is the absolute leader in China’s digital music scene with a market share of over 60%. Even after taking into account users gained via investment in NetEase Cloud Music, Baidu’s share remains in single digits.

China’s tech giants battle for smart speaker supremacy as price war rages on

Besides music and other content, e-commerce, smart home, digital payments, gaming, and local lifestyle services are potential areas where smart speakers can be monetized. In all of these areas, competitors are in a much better position to leverage their smart speaker market share. Alibaba can naturally link its smart speakers with e-commerce. It also has a robust content platform with Alibaba Music and Alibaba Literature. All of these help to reinforce Alibaba’s digital payment business. Tencent also has colossal gaming, literature, music, and online payment businesses that can be bolstered by smart speakers. This is not true for Baidu, which is weak in each of these areas.

Signifying nothing

Combined, Baidu could suffer losses of several billion dollars on new AI projects. This is significant as a ratio to its $14.8 billion revenue and operating income of $2 billion for last year. When it comes to investing in AI, Baidu has followed the footsteps of Alphabet whose revenue and profit are nearly 10 times as large. But the Chinese search engine can’t afford to up the ante on this scale.

The hope, of course, is that betting big on AI will redeem the struggling giant. But the chances of Baidu succeeding in its AI strategy within the next five years is low. Investors should remain highly cautious of Baidu’s bet on AI and its potential outcome.

National artificial intelligence (AI) plans, including those drafted by China, should promote international collaboration, not just permit it, according to Tom Mitchell, former dean of Carnegie Mellon University’s computer science school.

Why it matters: Mitchell, known as the father of machine learning, was speaking at the opening ceremony of the World Artificial Intelligence Conference (WAIC) in Shanghai on Thursday.

• Shanghai is attempting to reimagine itself as an international AI capital, and for the city, the WAIC event forms a major part of this push.
• China hopes to become a world leader in AI in the next ten years and has laid out plans to meet this goal.

“What I think these national strategies need is a distinction that says for win-win applications the rational strategy for every country is not just to allow collaboration but actually to promote it. And to find ways to, for example, share medical data internationally, and share algorithms and the hard engineering work.”

—Tom Mitchell, at WAIC on Thursday

Details: Mitchell said that AI applications in healthcare, education, and smart cities could benefit from researchers in different countries working together.

• Collaboration on AI could improve the general quality of life, as working together could accelerate development, Mitchell said.
• Mitchell hopes Chinese policymakers will encourage collaboration into their future AI plans.

Context: China has laid out goals that policymakers hope will make the country an AI trailblazer by 2030.

• The country is currently working to catch up with the US in its AI capabilities. According to numerous reports, China falls behind in terms of talent and hardware, but leads in deployment and data.
• China has strict rules governing the transfer of personal data abroad, which becomes ever more sensitive when the data is medical in nature.

Chinese policies governing civil-military fusion could limit the country’s plans to become a world leader in artificial intelligence (AI), according to a new report, as distrust of companies linked to the Chinese government grows amid US-China tensions.

Why it matters: China has set ambitious goals to become a world leader in AI by 2030 in order to move the country up the industrial value chain.

• The country currently lags behind the US in AI development as it lacks the necessary talent and hardware prowess to take the top spot.
• Washington is concerned that the close collaboration between China’s military and private sector, which in turn works with US companies, could give China a leg up in a global arms race.

“‘Civil-military’ integration makes it harder for [China’s]  firms to succeed in the global market because such policies foster distrust in other societies. A lack of trust will hinder Chinese firms’ ability to acquire significant global market share outside of nations that are taking part in China’s subsidized Digital Silk Road initiatives.”

—US-based think tank Center for Data Innovation (CDI) wrote in a report

Details: China has shown interest in the military applications of AI, along with the US and other countries. While China has voiced support for a ban on the use of autonomous weapons at the United Nations, it still supports developing so-called “killer robots.”

• CDI’s report, published on Monday, ranks the US, China, and the European Union on their AI abilities. The document grades capabilities according to talent, research, data, and hardware, among others.
• CDI said that China still lags behind the US in AI, but it’s catching up fast.
• Despite China’s huge population, it’s still short on talent, the report said. This shortage extends to the chipmaking sector, which the AI industry is reliant on to provide computing power.
• The quality of research papers in China is also an issue. The country ranks below the US and the EU.

Context: A number of companies, both foreign and Chinese, have faced international repercussions and censure for their work in China and with the country’s government.

• The US put Chinese telecommunications giant Huawei on a trade blacklist earlier this year over national security concerns.
• Meanwhile, members of the US military and figures from the tech sector have called out Google, which as an AI research center in Beijing, for its work in China, saying that it benefits the Chinese military. Peter Thiel, venture capitalist and Facebook board member went as far as calling Google’s China links “treasonous,” giving no evidence.

Search giant Baidu is no longer one of China’s five most valuable companies, as the company struggles to keep up with competitors that encroach further into its primary markets, Bloomberg reports.

Why it matters: Baidu has seen increasing competition from rivals Tencent and Bytedance, which are enticing advertisers and users with their short video and social media apps.

• In May, Baidu reported its first quarterly loss since it listed in 2005.
• The company was previously knocked out of the top three in market value by e-commerce giant JD and lifestyle services company Meituan.

Details: Internet giant NetEase overtook Baidu in market value after posting its earnings last week. Baidu’s share price has fallen by 40% this year, while NetEase, China’s second-largest gaming company, gained 11%.

• Baidu has lost $66 billion in capitalization since May 2018, Bloomberg reported.
• The company is part of the BAT triumvirate, which also includes Alibaba and Tencent. However, Bytedance is quickly taking Baidu’s place in the minds and on the phones of internet users in China.

Context: Baidu has commanded the internet search market in China since Google’s departure in 2010. However, the company is now facing competition from upstart Bytedance, which launched its own search engine this month.

• Baidu has long been criticized for the quality of its search results.
• The company has been accused of promoting results from its own platforms and hosting ads for questionable healthcare services.
• Last month, the company apologized after an employee posted a fake message on its news aggregator in which they claimed to be the father of a girl who went missing and whose body was later found in mid-July.

Venture capitalist and Facebook board member Peter Thiel has reiterated his stance on Google’s presence in China, saying that it is “unprecedented” for a company to refuse contracts with the US military while seeking greater interaction with China.

Why it matters: Google has faced heightened scrutiny for its presence in China following revelations that it was working on a censored search engine to comply with Chinese laws.

• The discussion has expanded to encompass questions about whether the company’s presence in China is benefiting the country’s military.
• Thiel previously called for the FBI and CIA to investigate whether the Chinese government has infiltrated Google.

Details: Thiel claimed that even if Google isn’t working with the Chinese military, the technology the company develops in the country “gets handed on.” Thiel, who is also founder of controversial data-mining company Palantir, made the comments in an interview with Fox News on Sunday.

• Google has repeatedly denied any military involvement in China.
• Thiel stopped short of referring to the company’s actions as treachery, despite previously calling Google’s work in China “treasonous.”
• Thiel said that Google had decided not to renew a contract with the US government while simultaneously pursuing its interests in China, a country he said is a geopolitical rival.

Context: Google’s AI research lab in Beijing has become a major point of contention, as critics say draw attention to the possible military applications of the technology the search giant develops in China.

• Despite not offering any significant consumer-facing services in the country, Google has several businesses in China, including ad sales.
• A Google executive claimed in July that the company had “terminated” the censored search engine it was developing for the Chinese market.
• Thiel has not been immune to controversy. Palantir, which he originally founded in 2003, has been accused of racist hiring practices and condemned for its work with US immigration authorities on deportations.

A research team at Baidu’s video platform iQiyi has released a new facial recognition dataset named iCartoonFace, Synced Review reports. The researchers say the anime-based dataset can apply to recognition research, cartoon person modeling, and image classification.

Why it matters: For China’s competitive streaming industry, the use of innovative artificial intelligence applications can prove valuable. The new dataset demonstrates iQiyi’s efforts to push boundaries in AI research as it looks to improve its animated content offering for users.

Details: iCartoonFace comprises over 68,000 annotated images with 2,639 identities from 739 anime and cartoon albums found online and in iQiyi’s video library.

• The researchers also proposed a new “dataset fusion” method to improve performance in cartoon-based facial recognition algorithms.
• While the new technique is considered “state-of-the-art,” it is not nearly as accurate as human-based techniques.

Context: iQiyi is China’s largest streaming site after crossing the 100-million subscriber mark in June. Along with competitors like Tencent Video and Bilibili, the firm is investing in animated content including feature length movies.

• The platform already leverages AI in a variety of ways to serve recommendations, cast actors and actresses and create customized promotional content.
• The appetite for anime in China remains strong, as evidenced by the recent success of movie Nezha, which became the country’s highest-grossing animation since its premiere on July 26.
• The country’s anime market has grown substantially in recent years and is expected to reach RMB 200 billion ($28.4 billion) this year.

Baidu, Alibaba, and Tencent (BAT) are the top three investors in AI startups among large Chinese companies. Most of such funding goes to those focused on application layers or algorithms. While the tech giant trio has only made strategic investments so far, the field has also seen a stream of government and VC funding in recent times. Baidu and Alibaba have also moved into chip design themselves.

These new chip firms generally fit into three main categories—pure fabless semiconductor startups (which outsource manufacturing), algorithm startups that are moving into chip design, and internet firms that are exploring design too.

In total, I can think of at least 20 Chinese companies that fit into these categories, all of which are designing AI chips. I am sure there are a lot more of them out there in the vast industrial landscape. Traditional fabless design companies in China are also increasingly designing their own AI chips, making the market ever-more competitive.

It isn’t just China though where we see non-traditional chip companies like Baidu and Alibaba moving into design. In the US, countless listed giants, including Tesla, Google, Amazon, Microsoft, Facebook, and Apple, are designing their own AI chips, specific to their required applications. How can AI chip startups from both China and the US alike expect to live long when they face competition from behemoths that boast such technical and economic strength?

This vertical integration is not just a threat to startups but also to traditional chip designers that previously considered internet companies their customers. This dynamic will shape the industry in both regions for years to come.

These startups have moved quickly from burning VC cash to coming up with architectural innovations, and now they are tasked with actually finding customers in a very competitive market. Industry voices say it is becoming increasingly difficult for them to differentiate themselves and any start-up late to the party will struggle to bring anything unique, new, or different to the table.

Those that really do innovate and pursue emerging technologies like analog computing, in-memory computing, and neuromorphic computing warrant attention but have to deal with a much longer path to get a commercial product to market.

‘Chinese’ chips

From a hardware design perspective, these Chinese AI chips aren’t necessarily all that Chinese. As most of the companies work to tight deadlines, they tend to license an IP rather than develop in-house. The majority with which I have spoken will gain authorization to use networks that link design parts together, known as on-chip interconnect. They also rely on development and debug tools from companies like Lauterbach. That’s not to mention that the higher-end chips mostly use TSMC for fabrication. Of course, a large proportion of designs also use Arm cores. Some I have spoken with have developed their own GPU or custom neural network core, but the majority have neither the time nor luxury to do so. The bottom line is VCs want to see returns fast!

Returns come much faster in other industries. With the typical design process for a semiconductor taking 18 months and requiring tens of millions of dollars to finance just the first few designs, there is a real need to seek out customers from the get-go.

Some more experienced chip-focused VCs in the country understand this and are more patient, but we can expect to see some of these startups die off over the next couple of years as the pressure from VCs builds. The sector will consolidate for other reasons as well though. Industry mainstays are expected to snap up those new firms that look the most promising. The trend has been seen already when America’s Xilinx, the world’s leading designer, and supplier of programmable logic devices, acquired Beijing chip unicorn DeePhi last year. Other players will resort to acquisitions as a means to enter the market, as was seen when Alibaba took over C-Sky.

At this point, I think it’s safe to say the necessity to have dedicated AI hardware is here to stay. Most chips have some kind of ‘AI’ functionality, and this will only increase with time. Sure, some design teams will have to decide how much effort they dedicate to AI, but given that almost every major player is now involved, it is difficult to see such functions not becoming critical features of most future designs.

For China’s AI chip startups, it is a balancing act of getting to market first, being innovative, and having patient investors. The most successful will be acquired or secure more money, and the lame will die. The flipside for them is that AI and semiconductors are two areas of focus for the Chinese government in its push for technological independence. Those able to combine these verticals will garner a lot of attention. Despite this, the market is still approaching saturation, and only those that strike the right balance will survive.

China Tech Investor is a weekly look at China’s tech companies through the lens of investment. Each week, hosts Elliott Zaagman and James Hull go through their watch list of publicly listed tech companies and also interview experts on issues affecting the macroeconomy and the stock prices of China’s tech companies.

Make sure you don’t miss anything. Check out our lineup of China tech podcasts.

Can’t see the player? Check us out on iTunes or Spotify!

In this episode of the China Tech Investor Podcast powered by TechNode, James and Elliott chat with Matt Sheehan, Fellow at Macro Polo, the in-house think tank of the Paulson Institute. Matt discusses some of his recent research, in which he analyses how US and Chinese firms stack up in various dimensions of data. He also shares his findings on the battle for AI engineers, and how each country’s AI future may rely on Chinese talent, on both sides of the Pacific.

Also, be sure to check out Matt’s upcoming book: The Transpacific Experiment: How China and California Collaborate and Compete for Our Future, out August 13th.

Please note, the hosts may have interest in some of the stocks discussed. The discussion should not be construed as investment advice or a solicitation of services.

Watchlist:

• Tencent
• Alibaba
• Baidu
• iQiyi
• Xiaomi
• JD.com
• Pinduoduo
• Meituan-Dianping

Guests:

•  Matt Sheehan – @mattsheehan88

Hosts:

• Elliott Zaagman – @elliottzaagman
• James Hull – @jameshullx

Editor

• Peter Isachenko

Podcast information:

• iTunes
• Spotify
• RSS Feed
• Music: “Hey Ho” by Steve Jackson, Royalty Free Music

Baidu is investigating how online search results for a line of political thought attributed to President Xi Jinping guided users to a video clip appearing to ridicule the Chinese leader, Bloomberg reported.

Why it matters: Baidu has long faced scrutiny for the quality of its search results. The company has been accused of promoting content from its own platforms and hosting questionable ads for healthcare services.

• The latest incident represents a major faux pas, as China’s leaders typically don’t take well to mockery.
• Baidu, like other tech companies operating in the country, is required to filter politically sensitive topics.

Baidu declined to comment when reached by TechNode on Friday.

Details: Baidu users looking for information about Xi’s “Four Greats,” a doctrine developed by the Chinese leader, were directed to results that included a video explaining the Chinese phrase for “tooting your own horn.”

• Baidu had fixed the issue by Tuesday evening, but is investigating how the two topics were linked, Bloomberg sources said.
• Some users circulated the search results privately on social media.

Context: Content platforms in China are required to censor sensitive text and images in order to escape government censure. Failure to abide by content-related laws can result in a company having to suspend its services or pay fines, or end up having its business license revoked.

• Baidu has faced increasing pushback for the content across its platforms. In 2016, a university student died of cancer after receiving ineffective treatment he had found through prominently placed ads in Baidu search results.
• More recently, the company apologized after posting a fake message in which the author claimed to be the father of a girl who went missing and whose body was later found in mid-July.

Artificial intelligence (AI) unicorn Sensetime is setting up a research center in the United Arab Emirates, with plans to hire hundreds of employees that will focus on AI in Europe, the Middle East, and Africa (EMEA).

Why it matters: While Sensetime has a significant presence in Asia, the Abu Dhabi-based research and development (R&D) center will mark the company’s first presence in the EMEA region.

• As the world attempts to rely less on fossil fuels, Gulf states are looking for new ways to stimulate their economies. Sensetime is setting up the R&D center with support from the Abu Dhabi Investment Office (ADIO), which aims to promote business in the emirate.

“Abu Dhabi offers us the ability to focus on innovation in key sectors, opens up a new customer base and importantly, a new talent pool.”

 —Xu Li, Sensetime co-founder and CEO in a statement

Details: The research center will develop AI for industries including healthcare, remote sensing, and education, according to Sensetime.  The company aims to provide support to regional industries and governments using AI.

• ADIO said it will provide Sensetime with a “comprehensive support package,” though no further details were given.
• The company says that Abu Dhabi gives it access to a greater pool of expertise, along with a larger customer base.
• Abu Dhabi’s government hopes that its partnership with Sensetime will aid it in developing its knowledge-based economy.

Context: Beijing-based Sensetime is one of the most valuable AI startups in the world, and already has offices in China, Hong Kong, Singapore, and Japan.

• The company is developing technology for autonomous driving, augmented reality, and medicine.
• Sensetime’s main focus is image recognition technology.
• The company has faced scrutiny for selling its facial recognition technology to Chinese law enforcement.
• It is one of many that has built up an empire on the back of China’s goal of omnipresent surveillance, with around 40% of the AI firm’s revenue coming from Chinese government contracts.

 If you can’t see the YouTube player above, try watching here instead.

Chinese innovation investment has continued to shrink as investor fundraising has cooled further this year. Only 271 private equity and venture capital firms in the country raised funds in the first half, down by over half compared with a year ago.

Still, given that a number of VCs raised money in 2018, Jixun Foo, managing partner at GGV Capital, believes the problem is not a lack of money, but where money goes. “There needs to be new innovations that drive new capital deployment,” Foo said at the recent RISE conference in Hong Kong.

Foo honed in on China’s mobility sector, in which GGV has solid experience as an early backer of major player Hellobike.

In the space of just a few years, China’s bike-sharing sector has boomed. The industry still exhibits great growth potential with demand remaining strong among the country’s 1.4 billion people. Mobility players are now also focused on a new race to provide rental services for electric two-wheelers. Hellobike is one of the early movers, having rolled out shared e-scooters back in September 2017 when ofo and Mobike were still battling it out in the shared-bike market.

The Alibaba-backed company took another step forward in June this year, inking a RMB 1 billion ($145 million) deal with Ant Financial and CATL, the country’s largest battery manufacturer, to install battery-swapping stations nationwide for e-scooters. Ride-hailing giant Didi quickly followed suit, forming a two-wheeler business group the same month as it vies for market share.

“We believe China’s bike market goes very deep and is still growing,” (our translation) Fischer Chen, Hellobike’s chief financial officer, said at RISE. With about 250 million two-wheeler motorists nationwide, there are 700 million e-bike rides happening each day in the country, triple that of shared bikes, the company estimated.

China’s bike-sharing bubble has burst with dozens of players going bankrupt over the past years as funding dried up. The market cooled as authorities banned operators from putting additional cycles into circulation on the streets of key cities in late 2017. National technical standards on electric bikes followed and took effect in April this year.

In an interview with TechNode at RISE, Foo maintains that Hellobike could actually benefit from government regulation in terms of its technology and product capabilities.

Unlike ride-hailing, which is a serviced dependent on human drivers, bike-sharing is a business that basically relies on hardware, Foo said. This means it is more suitable for management using technology and rules. Some typical examples include locating bikes more accurately using IoT and educating users more effectively with regulations. One of the key issues is the efficient operation of the bikes, he added.

The Chinese short ride market, populated by shared bikes and e-scooter players, has undergone some key reshuffling. Ofo, once a pioneer in the bike-sharing boom, is now on verge of bankruptcy amid mounting debts and massive layoffs. Mobike has also scaled back expansions since Meituan took over. The city services giant posted an RMB 4.55 billion loss last year after the acquisition. Chen claims that Hellobike has snared more than 60% share of the bike-rental market and for the e-bikes, the share is even higher at around 80%.

Return to rationality

In an interview with Chinese media earlier this year, Foo said as investors have returned to a more rational approach and the Chinese investment market is expected to see a higher capital efficiency over the next couple of years.

Efficiency is a constant area of focus throughout GGV’s investment portfolio. Hellobike has broken even in more than 100 domestic cities, CEO Yang Lei announced last October. The average operation cost for each blue and white bike is only RMB 0.3, while other players spend over RMB 1 to keep them in action.

Another GGV-invested company Xpeng Motors claimed a “much higher capital efficiency” compared with rivals, as the NEV startup focuses more on the mid-range market rather than luxury models. The recent nosedive in Nio’s stock price “is a good lesson for the rest of us… to try to be more efficient and more sustainable,” said Xpeng President Brian Gu at RISE. The company, a top seller among China’s EV players, claims it probably only needs to use a quarter of its capital to hit the same shipment numbers as Nio.

Looking forward

Foo maintains that the next wave of innovation is also on the way with the mass adoption of artificial intelligence and 5G across industries like logistics, automobiles, and healthcare.

“Last year we saw a number of IPOs and some of them didn’t do well, but things always go in cycles,” Foo said. “We see short-form videos from 3G to 4G, what will come next with 5G?” The venture capital firm is betting on mobility, electric vehicles, and smart cities going forward. It will invest more than one-third of its $1.88 billion of funding secured last year in the sectors.

With contributions from Wei Sheng.

US-based electric vehicle (EV) maker Tesla is looking to Apple to help build its case against a former employee it suspects stole self-driving technology before defecting to a Chinese rival, according to court documents.

Why it matters: Cao Guangzhi, a former self-driving engineer at Tesla, confirmed in a filing earlier this week that he had uploaded Tesla’s Autopilot source code to his personal iCloud account, which is run by Apple, before leaving the company.

• The EV maker has now subpoenaed documents from Apple to aid in its investigation, according to a filing from last week.
• Cao went on to join XMotors.ai, the US-based research division of Chinese EV maker Xpeng, following his departure from Tesla.
• Apple suspects an ex-employee who also joined XMotors stole some of its autonomous driving technology.

Xpeng was not immediately available for comment when reached by TechNode on Friday.

Details: Tesla filed the case against its former engineer in March, accusing him of uploading in excess of 300,000 files as well as source code to his personal cloud storage account before quitting in January.

• At the same time, the company accused four other former employees of leaking trade secrets to Silicon Valley-based autonomous driving startup Zoox.
• Cao has denied any misconduct but admits that he did not disclose that he had copied the files prior to his departure.
• Tesla didn’t inquire about any confidential materials or information regarding trade secrets when the engineer left the company, Cao’s lawyers said.
• Cao has already provided Tesla with emails from his Gmail account.
• Neither Apple nor Tesla have accused Xpeng of wrongdoing.

Context: The US Justice Department last year brought charges against a former Apple employee who it suspects stole technology from the US tech giant before joining XMotors. Zhang Xiaolang, the defendant in the case, pleaded not guilty in July last year.

• The accusations came amid increased trade tensions between China and the US, in which intellectual property theft is central to US complaints.
• China is pushing its capabilities in developing self-driving cars. By 2030, the country aims to become a world leader in artificial intelligence—the technology that underpins self-driving cars.

Artificial intelligence (AI) can ease the strain on China’s overburdened healthcare industry while increasing access to higher quality health services, said Ericson Chan, CEO of Pingan Technology.

Why it matters: Just 8% of China’s hospitals have received a triple-A grade, a classification that is reserved the specialized healthcare facilities. According to Chan, these hospitals service 50% of all patients in the country.

• The majority of China’s premier hospitals are found in large cities like Beijing and Shanghai, creating an unequal distribution of health services.

“You need to wait for more than three hours before you can see a doctor, and the consultations are no more than seven or eight minutes.” —Ericson Chan, CEO of Pingan Technology, told CNBC. 

Details: AI could help to alleviate some of these problems, as well as the issue of overworked doctors. China currently employs around two physicians per 1,000 patients. Nonetheless, in order for the technology to be effective, startups and tech companies need to understand the bottlenecks that businesses encounter, Chan said.

• AI could address issues beyond efficiency. In some cases, algorithms have been able to outperform some physicians when diagnosing certain diseases.
• Researchers from China and the US earlier this year trained an AI using electronic records from 1.3 million patient visits to a hospital in the southern Chinese province of Guangdong. It was able to meet the number of correct diagnoses and even beat those of junior physicians for ailments ranging from asthma to sinusitis.

Context: Pingan’s focus is applying AI in various industries including healthcare, finance, and smart cities. The company said it has developed a system that can predict how likely a patient is to suffer from a chronic illness before symptoms develop.

• Chan said that aside from diagnosis, Pingan’s AI can follow up with patients who are already receiving treatment.

While Amazon dominates the US smart speaker market with its Alexa-powered devices, competition in China remains fierce as some of the country’s biggest names in technology engage in an ongoing price war to vie for market share.

China supplanted the US as the largest market for smart speakers globally earlier this year. Sales hit 10.6 million units in the first quarter, a recent report from research firm Canalys shows. The country accounted for just over half of all sales of the devices globally in the first three months.

In the two years since their emergence in the Chinese market, smart speakers have evolved from niche gadgets into some of the most popular electronic devices among Chinese households. Their meteoric rise has been the result of a price war between some of the country’s biggest tech players that slashed average prices to under $20. While experts contend that the low pricing strategy may help educate consumers, there are concerns that the low-end devices could fail to meet users’ expectations.

Amazon introduced a new means of human-computer interaction in 2014 when the company launched its first Echo smart speaker, running Alexa. Apple co-founder Steve Wozniak lauded Echo as the “next big platform” in computing in an interview with CNBC in 2016.

After that, US tech giants piled in, with Facebook, Microsoft, Google, and Apple all announcing smart speaker products in the following years.

Major Chinese players like Alibaba, Baidu, and Xiaomi, joined the party in mid-2017. Sales skyrocketed from a mere 60,000 units in 2016 to 9 million last year, making the country the largest market worldwide, according to a report by German market research firm GFK.

Heavyweight players

Chinese e-commerce giant Alibaba launched its first smart speaker Tmall Genie in July 2017, allowing users to order items from its Tmall premium shopping site from the comfort of their sofas.

Tmall Genie, which uses the Hangzhou-based firm’s intelligent personal assistant service AliGenie,  initially went on sale for RMB 499 ($72.5). The price was low compared with the first generation Amazon Echo, which sold for around $180 but the market reaction was lackluster. Sales of AliGenie only numbered in the several tens of thousands, Chinese online media outlet Qdaily reported in November 2017. However, fortunes changed when Alibaba slashed prices for the annual Singles’ Day shopping festival on November 11 that year.

The company sold the speakers for RMB 99 via coupons and discounts, an 80% reduction of the original price. By the end of the 24-hour shopping festival, Tmall Genie had become the first smart speaker on the Chinese market to hit 1 million unit sales. This compares with total smart speaker sales in the country of 1.65 million units for the whole of 2017.

By March 2018, Alibaba had accumulated 2 million sales, a feat that had taken Amazon over one year to achieve with its Echo.

In the same month of Tmall Genie’s initial launch, Chinese smartphone maker Xiaomi joined the battle releasing the Mi AI Speaker. The product is capable of controlling other smart products around the home via voice assistant Xiao Ai.

The RMB 299 smart speaker hit the market in September 2017. Xiaomi said all stock sold out from the Beijing-based company’s online store 23 seconds after release, though it didn’t reveal how many were actually snapped up.

In March 2018, Xiaomi launched the Mi AI Speaker Mini, a low-end version of its original product, priced at RMB 169. For its annual Mi Fan Festival on April 3, Xiaomi cut the price to RMB 99, achieving parity with Alibaba’s Tmall Genie.

The low pricing strategy drove up sales significantly. Xiaomi said that more than 1 million users registered to buy the product during the festival.

The company sold 600,000 smart speakers in the first three months of 2018, reported Canalys. This number more than tripled to 2 million in the second quarter, installing the firm as China’s second-largest smart speaker vendor after Alibaba.

November 2017 marked the entry of a major new player when Baidu launched its Raven H product. The name derived from Raven Tech, the AI assistant startup that the search engine giant acquired in 2016.

Priced at RMB 1,699, Raven H was almost 10 times more expensive than other products available in the country and it failed to gain a foothold in the market. A report from The Information in June 2018 states that Baidu only manufactured 10,000 Raven H smart speakers, which is in sharp contrast to the several million units sold by both Xiaomi and Alibaba.

The failure led to Baidu pursuing the same strategy as Alibaba and Xiaomi. The firm unveiled its Little Fish smart speaker brand in March 2018, powered by its conversational AI assistant DuerOS. The device was priced at RMB 599 but was sold at a promotional price of RMB 299. In June 2018, the company launched a cheaper version of the device, which brought the price down to RMB 89.

The strategy paid off and Baidu became China’s third largest smart speaker maker after Alibaba and Xiaomi, with shipments of 1 million devices in the third quarter, Canalys data shows.

Price War

Behind the price war, there is a booming market for smart speakers. Canalys expects in April total units sold to hit 59.9 million at the end of this year, growth of 166% compared with the year-ago figure.

“The benefit of the price war is that it brings high market penetration in a short time, and it will be an effective measure for educating consumers,” Liu Hao, a researcher at the China Consumer Electronics Association, told TechNode.

“But the low pricing strategy won’t help the industry move forward; instead it will disappoint consumers’ expectations by selling them cheap stuff,” he said.

David Watkins, director at Strategy Analytics, told TechNode that Google and Amazon sold their devices at a very low cost to accelerate user number growth and now Chinese players are looking to replicate this success with even more aggressive promotions and deals.

“Given the sudden growth in sales in China over the last few quarters I would say that they are proving successful,” said Watkins.

Canalys predicts that smart speaker penetration will reach 13% this year though this is dependent on the ability of companies to win over new users, according to Watkins.

At the end of last year, Amazon and Google cut the prices of their smallest smart speaker products, the Echo Dot, and the Home Mini, to as little as $29 from $50 for the shopping season, reported Reuters.

The report indicated that the respective component costs of the two devices were around $31 and $26, respectively, and the figures didn’t include overheads, shipping, and other expenses, meaning they were likely sold at a loss.

Although Xiaomi aims to lower the threshold for smart speakers by providing products at low cost, the company does not aim to pursue loss leader pricing–sell at a loss to attract consumers, Tang Mu, the general manager of Xiaomi’s smart hardware division, told TechNode.

“Loss leader pricing is not a reasonable business model, because it means high expectations for future market volume, and the smart speaker market has not yet reached that stage,” Tang added.

Shared ambitions

Chen Xiaoliang, founder and CEO of SoundAI, a voice recognition technology provider for Xiaomi and Baidu’s smart speakers, told TechNode that the low-pricing strategy was a better way of marketing than advertising because smart speakers are still a new concept in China.

However, the country’s smart speaker makers are not as ambitious as their US counterparts because they have different purposes, he added.

“Chinese companies are selling smart speakers as a potentially profitable product, but they don’t envisage a scenario in which smart speakers represent the main focal point of people’s interaction with machines,” said Chen.

Voice interaction is the most natural way of asking and searching for answers, he said. “Amazon is ambitious because it is determined to dethrone Google as the top search engine, which would save it billions of dollars in search engine advertising.”

Watkins of Strategy Analytics still contends that Baidu and Alibaba share similar ambitions to Amazon and Google. “They see AI assistants as the next computing platform, and if they want to remain relevant then they must build a strong presence in that space.”

Companies in both China and the US alike have to evolve their products based on changes in consumers’ behavior as they moving away from desktop and touchscreen interfaces toward more direct voice-activated platforms, he said.

Baidu has been accused of copying Alibaba on the design of its latest smart speaker, a consumer electronics segment attracting intense competition among Chinese tech giants because it is seen as a gateway for smart home ecosystems.

Baidu’s latest smart speaker model, the Xiaodu Play, features several design details similar to those on the Tmall Genie R, which Alibaba released three months ago. Alibaba posted a comparison chart on its social media account late Wednesday displaying five points of resemblance between the two products, including pop art designs, color, gradient speaker holes, and concave buttons.

Li Jianye, design lead for the Tmall Genie, said the smart speaker has several blind holes along with the two microphone jacks for aesthetics, and an idea that was “copied exactly by Baidu.”

Baidu denied the allegations on Thursday, saying that it is leading the market due to its independent, well-grounded design philosophy, as well as research and development (R & D) strengths. “Don’t attack against us because we are now the number one,” (our translation) Jiemian reported citing a Baidu spokesperson.

Baidu did not respond to request for comment when contacted by TechNode on Friday.

Baidu unveiled three of its Xiaodu-branded smart speaker devices at the company’s annual developer conference earlier this week, all powered by DuerOS 5.0, the latest version of its voice assistant software.

Robin Li showed off the company’s recent achievements in speech recognition at the opening session, demonstrating how the artificial intelligence (AI)-based assistant interacts with users in a continuous dialogue without the need to activate the device with a wake word.

Rivalries among tech giants in the smart speaker market is rising in China, as the gadget is widely seen as an entry point into consumer smart home ecosystems. According to data from market research firm Canalys, Baidu shipped 3.3 million smart speakers in the first three months of 2019, surpassing Alibaba and ranking first in China. Alibaba had dominated the market following momentum gained from selling an excess of 1 million units during its Singles Day shopping event in November 2017.

Malaysia’s sovereign wealth fund appoints SenseTime founder to board as it seeks to expand tech investments – South China Morning Post

What happened: Khazanah Nasional, Malaysia’s sovereign wealth fund, has appointed the founder of artificial intelligence (AI) startup Sensetime as a board member to advise on tech investments. Tang Xiao’ou will advise on AI-related matters, marking the first time a foreign national has been appointed to such a position. Meanwhile, Lau Seng Yee, a Malaysian citizen and executive at Tencent, has also been appointed.

Why it’s important: Malaysia has shown increasing interest in Chinese technology, and Sensetime serves as a poster child of China’s tech development. Malaysian Prime Minister Mahathir Mohamad has previously visited the AI firm’s Beijing headquarters, and taken tours of both Alibaba’s and automaker Geely’s campuses during visits to China. Notably, the 93-year-old leader showed his support for telecommunications giant Huawei, saying that Malaysia would use its equipment “as much as possible” despite the Chinese company finding itself at the center of international scrutiny.

New ams/MEGVII Partnership for Plug-and-play solutions Enabling 3D Face Recognition in Any Smart Device – Businesswire

What happened: Chinese AI surveillance company Megvii will join forces with Austrian sensor manufacturer Ams AG to make what it claims will be the first off-the-shelf facial recognition solutions that don’t rely on smartphones. The standalone plug-and-play 3D suite, featuring Ams AG’s laser technology with Megvii’s facial algorithms, will focus on smart home, retail, security applications.

Why it’s important: Once referred to as “China’s rising AI star,” Megvii, which reached a $4 billion valuation in May, has recently entered choppier waters. The firm’s role in China’s state-level surveillance has attracted criticism from abroad since a March report stated Megvii was seeking an $800 million IPO listing. Many speculate that China’s surveillance players could soon face a US administration ban similar to that of Huawei. As a result, anonymous insider sources claim it is rethinking its listing altogether. Despite the turmoil, Megvii appears committed to partnerships and product development, this time with a European partner.

Malaysia is lagging behind regional peers in AI adoption, said Mohammad Radzi, executive director of G3 Global, during a Fireside Chat at ORIGIN, held during Malaysia Tech Week 2019. He said that the country should learn from China’s leapfrog success in the field.

Chinese AI company SenseTime is ramping up efforts to expand overseas presence with its latest deal to help build Malaysia’s first AI research park in a partnership with local AI company G3 Global. The park will also include an exhibition zone in which visitors can see AI in action.

“AI is burgeoning in Malaysia so we choose a partner who not only can grow local companies but at the same time spur AI innovation within Malaysia” said Radzi. This partnership aims to bring together and build an AI ecosystem in Malaysia. The AI park also plans to be an export centre, where Malaysian AI solutions can be exported to other countries, said Radzi.

“The AI park will be an area where visitors, not just industry professionals, can visit to experience AI solutions first hand,” said Radzi, describing a park where autonomous vehicles operate on the road and visitors can check into a condominium equipped with smart home system.

This partnership also plans to help build the country’s AI capability through an education curriculum, he said, noting that SenseTime has designed and developed an AI syllabus that is currently taught in schools across China.

Educating enterprises in Malaysia

Malaysia is trailing behind neighbouring countries in Southeast Asia on AI adoption. One contributing factor could be an AI knowledge gap among Malaysian enterprises,  Radzi said. “More could be done to educate enterprises on the potential of AI, such as improving productivity and increasing revenue generation,” said Radzi.

Size counts

China is huge, and when a proof-of-concept (POC) trial takes place, it is typically carried out on a large scale. “In contrast, kicking off a POC trial in Malaysia requires levels of authority clearance, and POC trial areas are just a fraction of what it’s like in China,” said Radzi.

AI works best if given large amounts of data sets, coupled with fast, iterative processing and intelligent algorithms. “China is able to advance its AI technologies at such a rapid pace due to the enormous amount of data collected through POC trials,” said Radzi.

China’s AI agenda advances as China throws state support behind AI development. According to the 13th Five-Year Plan timeframe (2016-2020), China has ambitions to transform itself into a superpower in science and technology. “It is great that the Chinese government is involved and actively spurring innovation within China. This could be the reason why China is advancing so rapidly in the AI race,” said Radzi.

MOST Launches “AI Voice Data Set” to Assist Chinese AI Language Technology – CTIMES

What happened: Taiwan’s Ministry of Science and Technology (MOST) released 400 hours of Chinese-language voice data to the public for use as training material for AI-powered voice applications. According to CTimes, the dataset includes self-recordings, as well as “data related to police and educational broadcasts.” It will be uploaded onto the National Center for High Performance Computing’s (NCHC) Data Market Platform, and is the first of multiple planned releases from MOST’s collection of 2,000 to 3,000 hours of voice data. 

Why it’s important: Voice recognition tech is one of the hottest subcategories of China’s rapidly growing AI industry, as evidenced by the country’s voice recognition unicorn and $9 billion-valued iFlyTek’s recent efforts to raise up to $350 million to invest in AI startups worldwide. And while iFlyTek and China’s other tech giants may hold a sizable share of the $55 billion voice recognition market, MOST’s data release can be particularly helpful to smaller players who lack large straining sets and are looking to improve the quality of their machine-learning processes.

Alibaba Cloud chief scientist Min Wanli on Saturday announced his resignation after six years with the e-commerce behemoth, the latest in a series events signaling slowing progress in smart transportation progress across China.

In a farewell letter obtained by Chinese media, Min said he had set up a venture capital firm to help fund the integration of cloud computing and artificial intelligence (AI) technologies into traditional industries such as manufacturing, agriculture, and healthcare.

A spokesman from Alibaba confirmed Min’s departure when contacted by TechNode on Monday. The handover was completed smoothly, and urban management and industrial solutions remain among key areas of focus for the company’s cloud business, he said.

After obtaining a doctorate in statistics from the University of Chicago and working as a data scientist at IBM and Google for nearly 10 years, Min joined Alibaba in September 2013 as its principal data scientist. He was tasked with developing data solutions for Alibaba’s online marketplaces Taobao and Tmall to drive sales and target users.

Min later served as vice president and chief scientist in Alibaba Cloud beginning mid-2017, leading the creation of ET City Brain, Alibaba’s cloud-powered and AI-driven urban traffic management system. The e-commerce and cloud giant in September launched the ET City Brain 2.0 in collaboration with the government of the eastern Chinese city Hangzhou. More than 1,300 traffic lights and 200 traffic officers were connected online to boost the city’s efficiency, according to the company.

Min’s departure casts a shadow on Alibaba’s ambitions in urban transportation digitization. In a recent report by Caixin, industry researchers expressed concern about the “slow progress” of intelligent transport system (ITS) construction nationwide. A researcher hired by the government said many government agencies have no idea how to collect useful information from data, or use data productively to improve traffic management.

Also, local governments are concerned about involving tech giants too much in urban management projects, and therefore most ITS projects across the country are slow to progress, Liu Liu, co-founder of Shanghai-based smart city solution provider CitoryTech told TechNode on Monday.

Min set up a company named North Summit Capital Management Limited with a registered capital of RMB 10 million ($1.45 million) in Shenzhen in late April, according to the company database website Qichacha.com. Chinese media reported it had received several hundred million dollars in its first round of funding.

This article was corrected to reflect North Summit Capital Management’s funding of several hundred million dollars, not $100 million.

This article by Eudora Wang originally appeared on China Money Network, the best data intelligence platform tracking China’s tech and venture capital markets (access requires subscription).

China has issued a series of principles on Monday to regulate the research and application of artificial intelligence (AI), in an attempt to ensure the “safe, controllable, and responsible use” of the technology amid rising privacy concerns in the country.

The principles highlight a theme of “developing responsible AI,” which include eight tenets, namely harmony and friendliness, fairness and justice, inclusiveness and sharing, respect for privacy, security and controllability, shared responsibility, open cooperation, and agile governance, said Xue Lan, director of China’s National Governance Committee for the New Generation Artificial Intelligence, cited by Chinese state-owned media Xinhua News Agency.

The committee, operating under the Ministry of Science and Technology (MOST), consists of a board of AI and public policy experts from various universities and research institutions to examine the impact of AI on laws, ethics, and society.

The principles stipulate that AI developers should conduct research in “a fair, inclusive, and open manner” that protects the interests of all related parties from developers to consumers. They also underscore other important issues in the domain, including privacy protection, international cooperation, responsible use of AI, and creating timely regulations to keep up with AI’s rapid development, according to the Chinese government website.

“AI technology is developing very fast and is changing everything in society, including economic structures, governance, national security, and even international relations,” said Xue. He said that AI technology has also raised many new concerns in domains like data privacy, machine ethics, AI safety and risks, and the misuse of AI technologies, referring to the misinformation initiated by “deep fake videos,” AI-manipulated fake footage that has become increasingly difficult for ordinary viewers to differentiate.

The move came after Beijing implemented a national AI plan in 2017, which predicts Chinese AI researchers leading an industry expected to be worth over US$150 billion by 2030. The huge amount of data being generated by Chinese citizens, however, has raised questions about how the Chinese state-owned and private companies collect, safeguard, and utilize the trillions of data points collected per day.

The Chinese metropolis of Shanghai is also drafting an AI industry development plan, the city’s vice mayor Wu Qing unveiled at a press conference on Tuesday. Wu said that the local government is also launching an AI-focused industry investment fund to further promote AI development after it welcomed the debut of China’s Nasdaq-style high-tech board last week.

A US senator proposed this month to ban companies based in China, North Korea, Russia, and other countries from an influential US government accuracy test of facial recognition technology, given that they consistently violate “internationally recognized human rights.” The test, known as “the Face Recognition Vendor Test (FRVT),” is widely recognized as the gold standard for evaluating the reliability of facial recognition software.

As artificial intelligence and fintech come knocking, half of Asia-Pacific finance professionals fear for their jobs – South China Morning Post

What happened: Half of the employees working in the finance industry in the Asia-Pacific region fear they will lose their jobs to artificial intelligence (AI), but the number of China’s financial professionals is expected to grow 26% in the next decade, according to a poll by the Chartered Financial Accountants Institute. The international association of investment professionals surveyed 3,832 members worldwide, a third of which live in Asia. Despite the rise of AI, China’s employment in financial services will grow at a rate second only to India, which will grow 33% over the same time frame.

Why it’s important: China is one of the world’s biggest investors in fintech, with tech giants like Tencent and unicorns like Lufax in the game, and also one of the biggest virtual payment markets. But the fintech industry has faced serious turmoil after a recent regulatory crackdown, particularly P2P lending. According to the report, AI does not threaten the overall growth of the sector. As China’s fintech industry evolves, the government is trying to increase its financial clout while attracting tech capital with a new Nasdaq-style exchange in Shanghai called the “STAR Market.”

China is pushing ahead in artificial intelligence, including applications to improve people’s health. But industry insiders say there are numerous challenges that stand in the way of automation in the healthcare sector.

“No matter how advanced our algorithms are, if we don’t have a high-quality data, the training results will be biased,” Chen Tong, chief health officer of IBM Greater China, said last week during a healthcare panel at CES Asia in Shanghai.

He added that good data should be well-structured but right now, there is too much “noise” in China’s healthcare databases, referring to unclear information contained in patient records. “We have hundreds of thousands of reports but only thousands are available for deep learning,” he said.

The industry faces a multitude of other issues, insiders say, including siloed information, the inability to transmit data effectively, and privacy concerns, all of which make implementation difficult.

John Gu, executive vice president and chief digital officer of Wuxi Nextcode, a genomics startup, said at the same event that algorithms, which need to be trained with a large amount of data, can’t make accurate decisions if hospitals are not willing to share information. Gu said healthcare facilities haven’t figure out a way to distribute the benefits that come with sharing data, as it’s hard to measure the contributions of individual hospitals.

Sharing data creates its own problems. “The privacy issue is still sensitive when it comes to the medical industry,” Ryan Zhang, CEO of oncology data platform LinkingMed, said at the event. “Hospitals are reluctant to share data since they don’t want to risk leaking personal information,” he added.

To ease hospitals’ concerns, China’s premier Li Keqiang last week signed regulations governing the management of human genetic information, aiming to clarify how to regulate the use of medical data, protect patient privacy, and preserve Chinese people’s genetic information.

According to the documents, China supports using human genetic information for scientific research, developing the pharmaceutical industry, and improving medical treatments. The use of some types of genetic information, including those that relate to “important families and special regions” (our translation), requires administrative approvals. No definition for this category has yet been given.

Zhang said that edge computing technology could be a good way to address privacy concerns by analyzing the data on a private server or device, rather than a public cloud.

Another problem is transfer speeds, particularly when it comes to big datasets for training algorithms in healthcare. “Gene data is far larger than all the data human beings produce on social media. To analyze it, a desktop [computer] is never enough, but 5G provides a solution by [efficiently] uploading the data to a more powerful server,” Chen said.

Ryan Li, head of Commercial Innovation & Alliance at Merk, one of the largest pharmaceutical companies in the world, said during a panel at the same event that 5G can help more patients from remote areas access the same medical resources as those in big cities such like Beijing and Shanghai.

The Chinese government issued its first 5G licenses for commercial use earlier this month. The move came shortly after Washington continued its offensive against Chinese telecommunication firm Huawei due to fears it could pose national security risks as a result of possible links to the Chinese government.

China’s smartphone companies including Huawei, ZTE and Xiaomi, and telecom operators like China Mobile, China Unicom, and China Telecom have all made public their ambitions to begin deployment of infrastructure and devices once licenses were granted.

“5G brings completely new possibilities for remote telemedicine, community healthcare, and remote surgery,” Zhang said.

On May 28, the Beijing Academy of Artificial Intelligence (BAAI) released the “Beijing AI Principles,” an outline to guide the research and development, implementation, and governance of AI. Endorsed by Peking University; Tsinghua University; the Chinese Academy of Sciences’ Institute of Automation and Institute of Computing Technology; and companies such as Baidu, Alibaba, and Tencent, the principles are the latest global entry in a long list of statements about what AI is and should be. On the surface, a bland and unsurprising take on AI ethics, the document actually pushes forward the global discussion on what AI should look like.

Bottom line: Say what you will about the current tension between the US and China, but the fact remains that the Western-built world order is slowly eroding and China is steadily filling in the gaps. While the principles were not written or endorsed by the State Council (the country’s chief administrative authority), BAAI is backed by the Ministry of Science and Technology which, very likely, approved the statement for publication.

Because it’s not a central government document, it does not carry the full weight of the leadership, but it is a step towards an official stance on ethical AI. It is the first public sector declaration of AI principles to come out of China, bringing Beijing into a conversation dominated by Western voices.

However, like other principles published globally before, it doesn’t do enough to address actual real world development and use cases. Instead of fluffy, feel-good utterances we can all agree on, the global AI community needs to go beyond just words and give us concrete examples of how AI can represent our highest values.

A brief timeline

• October 3, 2017: DeepMind, developers of AlphaGo and AlphaZero, releases “Ethics & Society Principles.”
• April 9, 2018: OpenAI, a non-profit founded by Elon Musk and Sam Altman, publishes its Charter, in English and Chinese
• April 16, 2018: UK’s House of Lords “AI in the UK: ready, willing and able?”
• May 18, 2018: Stanford launches Human-Centered Artificial Intelligence initiative.
• May 26, 2018: Robin Li, found and CEO of Baidu discusses Baidu’s AI principles (Chinese link) at a government data conference in Guizhou.
• June 7, 2018: Google publishes AI principles.
• September 2018: IBM releases “Everyday Ethics for Artificial Intelligence,” billed as a “practical guide for designers and developers.”
• October 12, 2018: Baidu joins the Partnership on AI, founded in 2016 by Google, Facebook, Amazon, IBM, and Microsoft.
• December 3, 2018: Jason Si, chairman of Tencent Research Institution, discusses Tencent’s AI philosophy at 2018 Peking-Stanford University Internet Law and Public Policy Conference.
• March 2019: Stanford upgrades HAI initiative to a university institute.
• May 22, 2019: OECD releases “Principles on AI.”
• May 28, 2019: BAAI releases Beijing AI Principles.

For a full list of statements and position papers on AI principles, check out Linking Artificial Intelligence Principles, a project started by one of the key authors of the Beijing AI Principles, Yi Zeng.

Chinese voices: As with AI research, there has been a substantial asymmetry in statements about AI ethics: many Chinese researchers and practitioners can speak English and know English language statements, but few outside China are aware of Tencent’s and Baidu’s. Neither have published their principles and most statements from their representatives have been made in Chinese for domestic consumption. By publishing in English, the Beijing AI Principles invites the rest of the world to have an open discussion with researchers, academics, and entrepreneurs in China how AI should be developed and implemented.

It’s different this time: While still lacking concrete steps for implementation, the Beijing Principles do have more substance than the recent OECD document. Consisting of five principles (all single sentences) and five suggestions to government (all single sentences), the OECD principles on AI are bland and mirror almost every other government document on the topic. The Beijing AI Principles, on the hand, outlines specific domains (research, use, and governance) and even acknowledges the risks of Artificial General Intelligence and Superintelligence.

The most comprehensive public discussion of AI principles out of China actually comes from Tencent, a company which has been criticized in China for lacking a comprehensive AI strategy. At the 2018 Peking-Stanford University Internet Law and Public Policy, Jason Si gave a speech outlining, in quite specific terms, how Tencent approaches AI development. Even he, however, doesn’t go into enough detail for any reasonable skeptic to believe in their principles without seeing it.

The dark side: The Beijing AI Principles were likely written in good faith, but it is hard to take them seriously given what we know about how China is currently using AI. Domestic policy doesn’t give much credence to either Western notions of limiting the power of the state over the individual or avoiding systems of variable rights by race.

As has been widely covered by Western media outlets, the Chinese government has been paying special attention to people of certain ethnicities living in certain regions. Afraid of losing control, the government has created mass surveillance systems in a bid to ensure stability in these areas.

In August 2018, researchers funded by the National Natural Science Foundation of China and China Education & Research Network Innovation Project published a paper in English detailing AI models that could identify ethnicity. China’s facial recognition companies have had a hard time with non-Han Chinese due to lack of data, but this goes far beyond just wanting to verify someone’s identity. Indeed, the application of AI here isn’t about who someone is, but rather which group they belong to for the purposes of profiling. But it’s not just here, but also in China’s smart city push that we can see AI’s power to monitor individual citizens.

The government’s Skynet and Sharp Eyes programs make it clear that policy prioritizes watching, tracking, and monitoring people inside China’s borders. Anytime you hear “smart city” in China, you can guarantee that video surveillance and facial recognition is a key component.

Almost every single tech giant is involved too.

Alibaba has long been developing their own smart city solutions in Hangzhou and in May 2019, brought that system to Shanghai, including 1,100 biometric facial recognition cameras. Huawei has been developing smart city hardware for many years, including smart cameras used for image and facial recognition. Baidu is helping to build Xiong’an, China’s high-tech powered “second capital.” PingAn, an insurance company that’s pivoted into tech, is working with the Hong Kong government to help build the city’s e-ID system.

End of the fluffy cycle

We’ve got enough [AI principles]. We need to start getting into the details. We need to get real about the conversations we should be having.

—Chris Byrd, research fellow at the Future of Humanity Institute at Oxford University

The bad news is that the Beijing AI Principles is yet another list of feel-good sentiments about problems in AI development. The good news is that almost everyone and their mother has published one by now: the territory is marked and it’s time for some real exploration. What all AI principles lack to-date are concrete and technical details about how to create artificial intelligence systems that actually adhere to our stated values. Not only is this possible (contrary to popular belief, non-engineers can understand how technology works if they take the time to ask), but becoming increasingly necessary as AI quickly becomes integrated into the fabric of our societies.

The Beijing AI Principles are written by a team with a good track record and there’s every reason to believe they mean what they say. AI ethics is another area where China could set the global agenda if they wanted to. There is a real gap currently in the discussion that China could try to bridge, not just because they want the power, but also because the country’s leadership recognizes that there are real problems that need to be solved. Say what you will about their domestic policies, with AI, like carbon, if you don’t have Chinese buy-in, you don’t really have an effective norm. The world should be eager to have China contributing to the conversation.

Read more:

• To make AI ethics work, we need Chinese companies to lead
• The Future of Surveillance
• Cross-sector collaboration is key to achieving ethical AI: Chris Byrd
• Linking Artificial Intelligence Principles

First Ever US/China Legal AI Partnership – LegalSifter + DocQBot – Artificial Lawyer

What happened: Pittsburgh-based LegalSifter is partnering with Chinese firm docQbot to provide AI-based contract review for Chinese import-export companies, according to trade publication Artificial Lawyer. The service will aid the companies in reviewing English-language cross-border contracts more efficiently. It will also produce commercial risk reports on annotated contracts in Chinese and rank individual risk issues identified in its review. LegalSifter CEO Kevin Miller said of the deal, “We are fortunate to be partnered with docQbot. We are a global company, and the Chinese market is one we intend to serve.”

Why it’s important: The service being provided by docQbot and LegalSifter seems to be mostly unique in China’s cross-border deal space. According to Artificial Lawyer, there are few “serious” AI platforms targeting this niche in the legal industry, and the potential market is at least as big as the estimated 5 million import-export businesses currently operating in China. What’s more, docQbot reports that the vast majority of these companies review English-language contracts without legal counsel, making them vulnerable to risks that would be otherwise avoidable with the appropriate tools.

Baidu is ramping up efforts to install its voice assistant DuerOS into cars in China as the government pushes for world leadership in intelligent connected vehicle technology by 2035.

DuerOS and the company’s internet of vehicle (IoV) platform, Apollo, has been installed on more than 300 car models from 60 auto brands to date, and will be installed in 200 more over the next two years, a Baidu executive said on Tuesday at this year’s CES Asia in Shanghai.

Li Zhenyu, vice president of Baidu’s Intelligent Driving Group, said it had partnered with a number of automakers, including Ford, Mercedes, BMW, and China’s Great Wall Motors. Baidu announced in January that DuerOS has reached over 200 million devices, but did not reveal the total number of vehicles with the software. Alibaba stated in August that its AliOS was installed in more than 700,000 vehicles, mostly SAIC-brand vehicles.

Powered by DuerOS, vehicles from Ford, Chery, and Great Wall Motors are now capable of intelligent navigation and entertainment services including content from video-streaming platform iQiyi and Himalaya FM, an audio content app. He Fei, an executive from telecom operator China Unicom, said the two companies will partner to develop favorable payment plans for in-vehicle applications in the future.

Tencent announced in November its launch plan for entirely voice-enabled WeChat services as part of its Tencent Auto Intelligence (TAI) software. However, the company later postponed the release to the end of this year to address public concern over the safety of its touch screen interface, which the company said it was stripping completely out. “It is a difficult task, especially for natural language processing,” (our translation) Tencent’s Pony Ma said publicly late last year, Xinhua.net reported.

As past of its 2035 goal, the Chinese government plans to complete wireless vehicle communication network (LTE-V2X) buildout, product standards, and regulatory guidelines by 2020, according to a strategic plan released by the state planning department, National Development and Reform Commission, in January 2018.

BatchNorm + Dropout = DNN Success! – Synced

What happened: A group of researchers from Tencent Technology, the Chinese University of Hong Kong, and Nankai University have proposed a new deep neural network (DNN) training method that aims to improve training efficiency in machine learning. By combining the Batch Normalization (BatchNorm) and Dropout techniques, the researchers were able to improve the stability of the training process and improve the convergence limit and speed of their network. According to their paper’s abstract on arXiv, the researchers’ “work is based on an excellent idea that whitening the inputs of neural networks can achieve a fast convergence speed.”

Why it’s important: Deep neural networks can become cumbersome when faced with large amounts of data and complex training models, so a new training method that improves efficiency is valuable to the development of this category of neural network. The new research also proposes a way for BatchNorm to refocus on improved whitening, which, according to Synced, “is a preprocessing technique that seeks to make data less correlated and standardize variance.” More recently, BatchNorm had shifted away from its goal of whitening due to increased computational cost.

Alibaba AI Labs, Alibaba Group’s AI research division, is partnering with automakers including Audi, Renault, and Honda to integrate its Tmall Genie Auto into certain models sold in China, the company announced at CES Asia in Shanghai on Tuesday.

Tmall Genie Auto, an artificial intelligence solution developed by Alibaba AI Labs, will offer a variety of voice-controlled information and services. The in-vehicle assistant will help drivers to identify nearby attractions and restaurants, book movie tickets, check the status of package deliveries, read children’s books, and order items on Alibaba’s retail platform.

In addition, car owners with a Tmall Genie-compatible devices, such as the Tmall Genie speaker and Tmall Genie mirror, will be able to monitor and control smart-home devices from their cars in the near future, according to the company. Drivers will be able to check on the temperature and lights, or turn on the heater and air conditioning at home from their vehicles.

“By providing AI technologies, including speech-recognition and Natural Language Processing, Tmall Genie Auto enables car users to access an extensive in-car infotainment portfolio by tapping into Alibaba’s rich content and service ecosystem,” said Miffy Chen, General Manager at Alibaba A.I. Labs said in an emailed statement.

Partnership with Alibaba will help Audi to offer more localized in-car voice assistant services to their Chinese customers, according to H.W. Vassen, senior director of Digitalization and NEV Development at Audi China.

The current deals are just the latest involving of Tmall Genie’s integration with automobiles. Since the launch of Tmall Genie Auto solution last April, Alibaba A.I. Labs has also partnered with auto brands including BMW and Volvo Cars.

Connected car technology uptake is expanding rapidly in China, with market growth expected to reach double digits from 2018 to 2023, according to data from market intelligence firm Netscribes. Compared with price or engine performance, Chinese consumers are increasingly emphasizing in-car technology when it comes to buying cars, and 40% are willing to change brands for better connectivity, according to the report.

Tech giants are quickly adapting to the changes in customer preferences. In addition to Alibaba, Baidu has its in-car operating system DuerOS and Tencent launched its TAI Smart Car System last year. In line with the trend, Tencent and Alibaba are developing in-vehicle versions for their respective WeChat and DingTalk platforms, popular instant massaging tools for personal and professional communications in China.

Confucius felt truly conflicted. He had watched humans spark the engine of the first train and boot the first computer, but nothing had given him such chills of uncertainty as watching them race to build an artificially intelligent decision-maker.

Of course, the Old Master was skeptical of their self-assurance that they could create such a machine, but he couldn’t help pondering its implications for the system of ethics he had created.