In the AI ​​race with the U.S., China is falling behind on a key weapon: its own OpenAI | Wilnesh News

NVIDIA The stock market’s rocket ride highlights the extent to which chip quality and availability will determine the winners in the artificial intelligence era. But there’s another dimension to measuring early leadership in the field. China is looking to produce its own chips or get more from Nvidia, but among dozens of Chinese tech giants and startups, OpenAI has yet to emerge as a dominant generation artificial intelligence contender.

Late in the game, China is looking to catch up with OpenAI’s lead in the broader U.S. artificial intelligence market built by tech giants Microsoft, alphabetical Google and Amazonand well-funded startups including Anthropic, which this week received a $2.7 billion cash infusion from Amazon.

The gap between the United States and its tech rival China is seen as huge in the fast-growing sector. “China’s leading companies are benchmarking against ChatGPT, which shows how far behind they are,” said Paul Triolo, senior vice president and head of technology policy in China at Dentons Global Advisors in Washington, D.C.

“There are not many companies that can support their own large-scale language models. It requires a lot of capital. Silicon Valley is definitely in the lead in this competition,” said Jenny Shaw, a partner at Leonis Capital, an AI venture capital firm in San Francisco.

The United States remains the largest investment market. According to data from CB Insights, global investment in artificial intelligence companies reached US$42.5 billion last year, nearly half of which came from the new generation of artificial intelligence upstarts. In the United States, venture capital and corporate investors drove artificial intelligence investment to $31 billion through 1,151 deals, led by huge spending by OpenAI, Anthropic and Inflection. In comparison, China had 68 deals worth $2 billion, down sharply from 377 deals worth $5.5 billion in 2022. The decline is partly due to U.S. restrictions on Chinese venture capital.

“China is at a big disadvantage in terms of building the basic models for a new generation of artificial intelligence,” said Ma Rui, an AI investor, investment group and co-founder of the podcast TechBuzz China.

But China lags behind in basic models, mainly OpenAI and Google’s Gemini. It is narrowing the gap by using Meta’s open source large-scale language model Llama 1. Triolo said that even if China’s competitors lag behind, they are still based on the American model. Make improvements.

“Many Chinese models are actually forks of Llama, and the consensus is that these forks are one to two years behind leading US company OpenAI and its video-to-text model Sora,” Ma said.

China does have the technical talent to play a role in the AI ​​competition in the coming years.

a new study Research by Marco Polo, a think tank under the Paulson Institute, shows that the United States has 60% of the top artificial intelligence institutions, and the United States remains the main destination for elite artificial intelligence talents, accounting for 57% of the total, while China accounts for 57%. 12%. But the study found that China leads the United States in a number of other indicators, including leading the United States in producing top artificial intelligence researchers (based on undergraduate degrees), at 47 percent and the United States lagging behind at 18 percent. Additionally, 38% of top AI researchers working at U.S. institutions are from China, compared with 37% from the United States

New entrants into China’s next-generation artificial intelligence market can also enable rapid mass adoption. Baidu’s ChatGPT competitor Ernie Bot was launched in August 2023 and reached 100 million users by the end of the year. Samsung plans to integrate Baidu’s Ernie AI into its new Galaxy S smartphones, while in another high-profile development involving U.S.-China relations, Apple is working with Baidu to supply the Chinese company’s Gen for the iPhone 16 AI technology for negotiation.

Leong said Baidu’s Ernie Bot model is considered the most advanced among the current list of artificial intelligence competitors.

Several other Chinese companies are also forging ahead with funding from major players in the domestic technology market. Large cloud companies such as Baidu and Alibaba, social media companies such as ByteDance and Tencent, technology companies such as SenseTime, iFlytek, Megvii Technology and Horizon Robotics, and research institutions are all assisting in this effort.

Moonshot AI, funded by Chinese e-commerce giant Alibaba and venture capital firm Sequoia Capital (formerly Sequoia China), is building large-scale language models that can handle long content inputs. At the same time, former Google China President Kai-Fu Lee developed an open source new generation artificial intelligence model 01.AI, funded by Alibaba and its company Innovation Works.

Although China has accelerated the development of its local chip industry and advanced artificial intelligence, its artificial intelligence development is restricted to a certain extent by the United States’ export restrictions on high-end artificial intelligence chips. This market is monopolized by Nvidia and has become a battle between China and the United States for technological supremacy. part of the new battlefield. United States and China.

Bernard Leong, founder and CEO of Singapore-based technology consultancy Analyze Asia, said: “Despite efforts to develop local solutions, Chinese artificial intelligence developers still rely heavily on foreign hardware, especially hardware from American companies. This is a problem in the current geopolitical situation. It’s a weakness in the political climate.”

Ongoing tensions between the United States and China over technological innovation and national security issues are leading to fragmentation in the development of a new generation of artificial intelligence, following the pattern of other influential technologies locked in a superpower technological arms race. Given regulations and bans on sensitive, cutting-edge technologies, the likely outcome is the emergence of two parallel ecosystems for a new generation of artificial intelligence, one in the United States and one in China. ChatGPT is blocked in China, and Baidu’s Ernie Bot can only be accessed in the United States with mainland Chinese mobile phone numbers. “American companies cannot enter China, and Chinese companies cannot enter the United States,” Xiao said.

U.S. Commerce Secretary Gina Raimondo said that the United States’ purpose in restricting the export of artificial intelligence chips is to prevent China from obtaining or producing advanced chips. As China focuses on local capabilities, Chinese companies Semiconductor Manufacturing International Corporation or Huawei could become alternatives to Nvidia. But if export controls deny these companies access to state-of-the-art manufacturing designs, the future of alternatives may be uncertain. Triolo pointed out that Huawei has recently developed a series of AI chips as a competitor to Nvidia.

China is a leader in applying artificial intelligence to certain areas such as computer vision. “The chip shortage is very important for training basic models because you need certain chips, but for applications, you don’t need those,” Ma said.

Triolo said the “real killer apps” for Gen AI will be among companies willing to spend money to leverage the technology as part of their business operations. Alibaba is committed to integrating artificial intelligence into its e-commerce ecosystem. Leong said Huawei, although it was more successful in competing with Apple’s iPhone in the consumer market last year, also has broader ambitions to use its in-house hardware to develop artificial intelligence for specific industries, including mining.

Research from Boston Consulting Group suggests it may be some time before the broader artificial intelligence market develops beyond technology. Of the 1,400 senior executives surveyed, 60% are waiting to see how AI regulations will be developed, while only 6% of companies have trained employees on AI tools.

Artificial intelligence and technology issues are front and center for China’s leadership, and after ChatGPT’s breakthrough, China will loosen its artificial intelligence guardrails in 2023 and then revise some measures.

Open source generation AI technology used by many Chinese developers can encourage global collaboration and lead to shared insights as AI advances, but Leong said open source also comes with ensuring the quality and safety of models and managing bias and potential Related issues include misuse of artificial intelligence.

“China wants to make sure that content doesn’t come flooding back. They also want their companies to take the lead and be willing to take drastic measures,” Triolo said.

Ethical and social issues have hindered the progress of a new generation of artificial intelligence in China and elsewhere, including the United States, as evidenced by the battle over control of the OpenAI mission. Leong said there is another factor that could slow the acceleration of AI in China: maintaining control over new generations of AI applications, especially in areas sensitive to national interests.