China’s race to dominate AI comes up against dependence on the US – 02/29/2024 – Tech

In November, a year after the launch of ChatGPT, a relatively unknown Chinese startup jumped to the top of a ranking measuring the abilities of open-source artificial intelligence systems.

The Chinese company 01.AI had only been around for eight months, but it had strong backers, a valuation of US$1 billion and was founded by a well-known investor and technologist, Kai-Fu Lee. In interviews, Lee presented his AI system as an alternative to Meta’s generative AI model called LLaMA.

There was just one twist: part of the technology in 01.AI’s system came from LLaMA. Lee’s startup built on Meta’s technology, training its system with new data to make it more powerful.

The situation highlights a situation that many in China openly admit. Even as the country races to develop generative AI, its companies are based almost entirely on systems created in the United States.

China now lags the United States by at least a year in generative AI, and may be falling further behind, according to several technology industry insiders and engineers in leadership roles.

This sets the stage for a new phase in the fierce technological competition between the two countries, compared by some to a cold war.

“Chinese companies are under tremendous pressure to keep up with U.S. innovations,” said Chris Nicholson, an investor at venture capital firm Page One Ventures who focuses on AI technologies. The launch of ChatGPT was “yet another Sputnik moment that China felt it needed to respond to.”

Jenny Xiao, partner at Leonis Capital, an investment fund that focuses on AI-powered companies, said the models Chinese companies build from scratch are “not very good,” leaving many of them often using “tweaked versions of models Westerners”.

She estimated that China was two to three years behind the United States in developing generative AI.

The race for leadership in AI has enormous implications. Advances in generative AI could unbalance global technological power, increasing people’s productivity, helping industries and leading to future innovations even as countries grapple with the risks of technology.

As Chinese companies look to catch up by turning to open-source AI models from the United States, Washington is in a difficult position. Although the US tried to slow down China’s advances by limiting the sale of microchips and restricting investments, it did not stop the practice of releasing software.

For China, the new dependence on AI systems from the United States — mainly Meta’s LLaMA — raises deeper questions about the country’s innovation model, which in recent decades has surprised many by revealing prominent companies such as Alibaba and ByteDance, owner of TikTok, despite authoritarian controls.

“When Chinese companies are using American open source technologies to upgrade themselves, the issues become very complicated — and embroiled in questions of national security and geopolitics,” said Oren Etzioni, a University of Washington professor who specializes in AI and founder of TrueMedia. .org, a non-profit organization that works to identify fake news in political campaigns.

In an emailed statement, Lee, founder of 01.AI, said his startup’s AI model was built on top of LLaMA just like “most other AI companies,” adding that the use of open source technologies is a standard practice.

He said his company trained its AI model from scratch, using its own data and algorithms. These were “the main determinants” of the “excellent performance” of 01.AI’s model, Lee said.

Nick Clegg, head of global affairs at Meta, said openly sharing the company’s AI models helped spread its values ​​and standards, and in turn helped ensure American leadership.

AI has long been a priority in China. After AI tool AlphaGo defeated two of the top players in the board game Go in 2016 and 2017, Chinese politicians laid out an ambitious plan to lead the world in technology by 2030. The government has promised billions to researchers and companies focused on AI.

When OpenAI launched ChatGPT in November 2022, many Chinese companies were being hurt by a regulatory crackdown from Beijing that discouraged experimentation without government approval.

Chinese technology companies were also subject to censorship rules designed to manage public opinion and silence opposition to the Chinese Communist Party.

Chinese companies with the resources to build a generative AI model faced a dilemma. If they created a chatbot that said something wrong, its creators would pay the price. And no one could be sure what might come out of a chatbot’s digital mouth.

“It’s simply not possible to eliminate all the problematic ways these systems can express themselves,” said Andrew Ng, a professor of computer science at Stanford University and a former executive at Baidu, the Chinese search giant.

Wang Changhu, former head of ByteDance’s AI lab, founded a company called AIsphere in Beijing last year to spearhead what he saw as the next big frontier in technology — video generation.

In November, the startup launched PixVerse, an AI-powered generator that can create videos from a text description.

“We move forward, building our models from scratch,” Wang said. “This gives us a significant advantage as true pioneers in the field of video generation.”

This advantage may have only lasted a few months. Last week, OpenAI unveiled Sora, an AI tool that turns a simple text prompt into videos that look like they were taken from a Hollywood movie. Sora went viral instantly.