In November, a year after ChatGPT's release, a relatively unknown Chinese startup jumped to the top of a leaderboard that assessed the capabilities of open-source artificial intelligence systems.

The Chinese company 01.AI was only eight months old, but had deep-pocketed backers and a $1 billion valuation, and was founded by a well-known investor and technologist, Kai-Fu Lee. In interviews, Mr Lee pitched his AI system as an alternative to options such as Meta's generative AI model, called LLaMA.

There was only one twist: some of the technology in 01.AI's system came from LLaMA. Mr Lee's start-up then built on Meta's technology and trained the system with new data to make it more powerful.

The situation is emblematic of a reality that many in China openly admit. As the country races to build generative AI, Chinese companies rely almost entirely on underlying systems from the United States. China is now at least a year behind the United States in generative AI and may fall further behind, according to more than a dozen tech industry insiders and leading engineers, paving the way for a new phase in the cutthroat technological competition between the two countries. which some compare to a cold war.

“Chinese companies are under enormous pressure to keep up with American innovations,” said Chris Nicholson, an investor at the venture capital firm Page One Ventures that focuses on AI technologies. The release of ChatGPT was “yet another Sputnik moment that China felt it had to respond to.”

Jenny Xiao, a partner at Leonis Capital, an investment firm focused on AI-powered companies, said the AI ​​models that Chinese companies build from scratch “are not very good,” leading many Chinese companies to often “refined versions”. of Western models.” She estimated that China is two to three years behind the United States in generative AI developments.

The battle for the primacy of AI has enormous consequences. Breakthroughs in generative AI could tilt the global technological balance of power, increase people's productivity, benefit industries and lead to future innovations, even as countries grapple with the risks of the technology.

As Chinese companies look to catch up by turning to open-source AI models from the United States, Washington finds itself in a tough spot. Even as the United States has tried to slow China's progress by restricting sales of microchips and reducing investment, it has not stopped the practice of openly releasing software to encourage adoption.

For China, the newfound dependence on AI systems from the United States – especially Meta's LLaMA – has fueled deeper questions about the country's innovation model, which has surprised many in recent decades by producing world-renowned companies like Alibaba and ByteDance despite authoritarianism control of Beijing.

“When Chinese companies use American open source technologies to catch up, the questions become very complicated – mired in issues of national security and geopolitics,” said Oren Etzioni, a professor at the University of Washington who specializes in AI and the founder of TrueMedia.org, a nonprofit that works to identify disinformation online in political campaigns.

In an emailed statement, Mr Lee, the founder of 01.AI, said his startup's AI model was built on 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. Those were “the key determinants” of the “excellent performance” of 01.AI's model, Mr Lee said.

Meta pointed to comments from Nick Clegg, who heads global business, in which he said openly sharing the company's AI models helped spread the company's values ​​and standards, and in turn helped securing American leadership.

(The New York Times has sued ChatGPT creator OpenAI and its partner, Microsoft, for copyright infringement of news content related to AI systems.)

AI has long been a priority in China. After the AI ​​tool AlphaGo defeated two top players of the board game Go in 2016 and 2017, Chinese policymakers have drawn up an ambitious plan to become the world's technology leader by 2030. The government has pledged billions to researchers and companies focusing on AI

When OpenAI released ChatGPT in November 2022, many Chinese companies were hampered by a crackdown from Beijing that discouraged experimentation without government approval. Chinese tech companies have also been burdened by censorship rules designed to control public opinion and dampen strong 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 the wrong thing, its creators would pay the price. And no one knew for sure what might fall out of a chatbot's digital mouth.

“It's just not possible to eliminate all the problematic ways these systems can manifest themselves,” said Andrew Ng, who teaches computer science at Stanford and was a former executive at Baidu, the Chinese search giant.

Chinese tech giants also grappled with new regulations dictating how AI models can be trained. The rules limit the datasets that can be used to train AI models and the applications that are acceptable, and also impose requirements on registering AI models with the government.

“It is both harder and riskier to innovate in generative AI within the current regulatory regime, which is still a moving target,” said Kevin Xu, the US-based founder of Interconnected Capital, a hedge fund that invests in AI -companies.

Technology investors in China have also pushed for quick turnarounds on AI, which has led to money flowing into easy-to-implement applications rather than more ambitious goals focused on basic research, says Yiran Chen, a John Cocke Distinguished Professor of Electrical and Computer Engineering. at Duke University. As much as 50 percent of China's AI investments have gone into computer vision technology, which is needed for surveillance, rather than building basic models for generative AI, he said.

Now Baidu, Alibaba, dairy company Mengniu and engineering firm TAL Education have all jumped into the generative AI race in China, prompting the Chinese media to coin the term “the battle of 100 models” to describe the madness.

Some have criticized the “free for all” as publicity stunts that add unnecessary competition. Last year, Baidu CEO Robin Li described in a panel discussion that having hundreds of AI base models is a waste.

“More resources need to be allocated to applications across industries, especially given the limitations of our computing power,” he said.

Success was elusive. When Baidu introduced its chatbot Ernie in March, it turned out that the 'live' demonstration was pre-recorded. Baidu's shares plunged 10 percent that day.

Despite the setback, Baidu remains one of China's few major efforts to build a foundational AI model from scratch. Others are led by Alibaba and Tencent, the Chinese tech giants, and by a start-up affiliated with Tsinghua University.

A Baidu spokesperson declined to comment.

US restrictions on sales of AI chips to China pose further challenges, as many such chips are needed in training generative AI models. Baidu and 01.AI, among others, have said they have stockpiled enough chips to continue operations for the foreseeable future.

There are some bright spots for China in AI, including in areas such as computer vision and autonomous vehicles. Some Chinese entrepreneurs also want to give the United States an edge with breakthroughs in other areas of generative AI

Wang Changhu, the 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 released PixVerse, an AI-powered generator that can create video from a text description.

“We went ahead and built our models from the ground up,” Mr. Wang said. “This gives us a significant lead as true pioneers in video generation.”

That lead 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 lifted from a Hollywood movie. Sora immediately went viral.

In November, a year after ChatGPT's release, a relatively unknown Chinese startup jumped to the top of a leaderboard that assessed the capabilities of open-source artificial intelligence systems.

The Chinese company 01.AI was only eight months old, but had deep-pocketed backers and a $1 billion valuation, and was founded by a well-known investor and technologist, Kai-Fu Lee. In interviews, Mr Lee pitched his AI system as an alternative to options such as Meta's generative AI model, called LLaMA.

There was only one twist: some of the technology in 01.AI's system came from LLaMA. Mr Lee's start-up then built on Meta's technology and trained the system with new data to make it more powerful.

The situation is emblematic of a reality that many in China openly admit. As the country races to build generative AI, Chinese companies rely almost entirely on underlying systems from the United States. China is now at least a year behind the United States in generative AI and may fall further behind, according to more than a dozen tech industry insiders and leading engineers, paving the way for a new phase in the cutthroat technological competition between the two countries. which some compare to a cold war.

“Chinese companies are under enormous pressure to keep up with American innovations,” said Chris Nicholson, an investor at the venture capital firm Page One Ventures that focuses on AI technologies. The release of ChatGPT was “yet another Sputnik moment that China felt it had to respond to.”

Jenny Xiao, a partner at Leonis Capital, an investment firm focused on AI-powered companies, said the AI ​​models that Chinese companies build from scratch “are not very good,” leading many Chinese companies to often “refined versions”. of Western models.” She estimated that China is two to three years behind the United States in generative AI developments.

The battle for the primacy of AI has enormous consequences. Breakthroughs in generative AI could tilt the global technological balance of power, increase people's productivity, benefit industries and lead to future innovations, even as countries grapple with the risks of the technology.

As Chinese companies look to catch up by turning to open-source AI models from the United States, Washington finds itself in a tough spot. Even as the United States has tried to slow China's progress by restricting sales of microchips and reducing investment, it has not stopped the practice of openly releasing software to encourage adoption.

For China, the newfound dependence on AI systems from the United States – especially Meta's LLaMA – has fueled deeper questions about the country's innovation model, which has surprised many in recent decades by producing world-renowned companies like Alibaba and ByteDance despite authoritarianism control of Beijing.

“When Chinese companies use American open source technologies to catch up, the questions become very complicated – mired in issues of national security and geopolitics,” said Oren Etzioni, a professor at the University of Washington who specializes in AI and the founder of TrueMedia.org, a nonprofit that works to identify disinformation online in political campaigns.

In an emailed statement, Mr Lee, the founder of 01.AI, said his startup's AI model was built on 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. Those were “the key determinants” of the “excellent performance” of 01.AI's model, Mr Lee said.

Meta pointed to comments from Nick Clegg, who heads global affairs: saying that openly sharing the company's AI models helped spread its values ​​and norms, and in turn helped secure American leadership.

(The New York Times has sued ChatGPT creator OpenAI and its partner, Microsoft, for copyright infringement of news content related to AI systems.)

AI has long been a priority in China. After the AI ​​tool AlphaGo defeated two top players of the board game Go in 2016 and 2017, Chinese policymakers have drawn up an ambitious plan to become the world's technology leader by 2030. The government has pledged billions to researchers and companies focusing on AI

When OpenAI released ChatGPT in November 2022, many Chinese companies were hampered by a crackdown from Beijing that discouraged experimentation without government approval. Chinese tech companies have also been burdened by censorship rules designed to control public opinion and dampen strong 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 the wrong thing, its creators would pay the price. And no one knew for sure what might fall out of a chatbot's digital mouth.

“It's just not possible to eliminate all the problematic ways these systems can manifest themselves,” said Andrew Ng, who teaches computer science at Stanford and was a former executive at Baidu, the Chinese search giant.

Chinese tech giants also grappled with new regulations dictating how AI models can be trained. The rules limit the datasets that can be used to train AI models and the applications that are acceptable, and also impose requirements on registering AI models with the government.

“It is both harder and riskier to innovate in generative AI within the current regulatory regime, which is still a moving target,” said Kevin Xu, the US-based founder of Interconnected Capital, a hedge fund that invests in AI -companies.

Technology investors in China have also pushed for quick turnarounds on AI, which has led to money flowing into easy-to-implement applications rather than more ambitious goals focused on basic research, says Yiran Chen, a John Cocke Distinguished Professor of Electrical and Computer Engineering. at Duke University. As much as 50 percent of China's AI investments have gone into computer vision technology, which is needed for surveillance, rather than building basic models for generative AI, he said.

Now Baidu, Alibaba, dairy company Mengniu and engineering firm TAL Education have all jumped into the generative AI race in China, prompting the Chinese media to coin the term “the battle of 100 models” to describe the madness.

Some have criticized the “free for all” as publicity stunts that add unnecessary competition. Last year, Baidu CEO Robin Li described in a panel discussion that having hundreds of AI base models is a waste.

“More resources need to be allocated to applications across industries, especially given the limitations of our computing power,” he said.

Success was elusive. When Baidu introduced its chatbot Ernie in March, it turned out that the 'live' demonstration was pre-recorded. Baidu's shares plunged 10 percent that day.

Despite the setback, Baidu remains one of China's few major efforts to build a foundational AI model from scratch. Others are led by Alibaba and Tencent, the Chinese tech giants, and by a start-up affiliated with Tsinghua University.

A Baidu spokesperson declined to comment.

US restrictions on sales of AI chips to China pose further challenges, as many such chips are needed in training generative AI models. Baidu and 01.AI, among others, have said they have stockpiled enough chips to continue operations for the foreseeable future.

There are some bright spots for China in AI, including in areas such as computer vision and autonomous vehicles. Some Chinese entrepreneurs also want to give the United States an edge with breakthroughs in other areas of generative AI

Wang Changhu, the 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 released PixVerse, an AI-powered generator that can create video from a text description.

“We went ahead and built our models from the ground up,” Mr. Wang said. “This gives us a significant lead as true pioneers in video generation.”

That lead 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 lifted from a Hollywood movie. Sora immediately went viral.

A White House spokesman said Thursday that the anti-satellite technology developed by Russia had not been deployed and posed “no immediate threat to anyone's safety.”

“We are not talking about a weapon that could be used to attack people or cause physical destruction here on Earth,” the spokesman, John F. Kirby, told reporters.

Mr. Kirby also pushed back on calls from Republicans for a broad release of intelligence about the weapon came to light on Wednesday in part after a cryptic message from Representative Michael R. Turner, Republican of Ohio and the chairman of the House Intelligence Committee.

After Mr. Turner's announcement, current and former officials briefed on the matter said Russia is likely making progress on a new space-based nuclear weapon designed to threaten America's vast satellite network.

President Biden receives regular briefings and the White House had “serious concerns” about releasing any information collected on the weapon, Mr. Kirby said. He added that the weapon was based in space and had no component that could be used to disarm or attack structures in space from the ground.

Mr. Kirby said the administration would advocate for declassification in a “strategic manner” and that Jake Sullivan, the president's national security adviser, would meet with members of the House leadership to brief them on the latest intelligence and its analysis by the government.