The United States and China are cooperating more closely on artificial intelligence than you might think

United States and China is, by many measures, a bitter competitor in this area artificial intelligencewhere companies compete to outdo each other Algorithms, Modelsand Specialized silicone. However, the world’s AI superpowers still cooperate to a surprising degree when it comes to cutting-edge research.

WIRED’s analysis of more than 5,000 AI research papers presented last month at the industry’s premier conference, Neural Information Processing Systems (Norebus), reveals a great deal of cooperation between American and Chinese laboratories.

The analysis found that 141 of the total 5,290 papers (about 3%) involved collaboration between authors affiliated with US institutions and those affiliated with Chinese institutions. US-China collaboration appears fairly steady as well, with 134 out of a total of 4,497 papers co-authored by institutions in both countries in 2024.

WIRED also looked at how algorithms and models developed in one country can be shared and adapted across the Pacific. The transformer architecture, developed by a team of researchers at Google and now widely used across industry, has been presented in 292 research papers with authors from Chinese institutions. Meta’s Llama family of models was a core component of the research presented in 106 of these papers. At the same time, The increasingly popular large language model Qwen From Chinese tech giant Alibaba, it appears in 63 research papers including authors from US organizations.

Jeffrey Ding, an assistant professor at George Washington University who tracks China’s AI scene, says he’s not surprised to see this level of teamwork. “Whether policymakers on both sides like it or not, the AI ​​systems of the United States and China are inextricably intertwined, and both benefit from this arrangement,” Ding says.

Of course, the analysis oversimplifies the degree to which the United States and China share ideas and talents. Many Chinese-born scholars study in the United States and form lifelong connections with colleagues.

“NeuroIPS itself is an example of international collaboration and a testament to its importance in our field,” Katherine Gorman, a spokeswoman for NeurIPS, said in a statement. “Collaboration between students and advisors often continues long after the student has left their university. You can see these kinds of signs of cross-field collaboration in many places, including professional networks and previous collaborators.”

the Latest issue of WIRED Explores the many ways in which China is shaping the current century. But with US politicians and tech executives using concerns about China’s rise as justification Trenching regulations and Fuel amazing investmentsOur analysis is a good reminder that the world’s two AI superpowers still have a lot to gain from working together.

A note on methodology

I used Codex, OpenAI’s code writing model, to help analyze the NeurIPS papers. After writing a script to download all the papers, I used the template to drill down into each one and do some analysis. This included the Codex Alimentarius Commission scripting a search for US and Chinese institutions in the author field for each paper.

The experiment provided a fascinating glimpse into the potential for programming models to automate useful routine tasks. There’s a lot of panic about AI replacing programming jobs, but that’s something I wouldn’t normally have the time or budget to build. I started writing scripts and asking Codex to modify them before asking Codex to perform the analysis itself. This includes the model importing Python libraries, testing different tools, and writing scripts before producing reports for me to examine. The process involved a fair amount of trial and error, and you have to be very careful, because AI models make surprisingly stupid mistakes even when they’re very smart. I had to make sure that each report included a way for me to review the results, and I checked as many as I could manually.

This is an edition of Will Knight Artificial Intelligence Lab Newsletter. Read previous newsletters here.