While some of the world’s most decorated mathematicians in the US and Europe have spent the past few months publicly warning that AI is corroding their field, China’s mathematical establishment just did the opposite: it organized.
On Sunday, leading Chinese mathematicians launched the Yau Laboratory Alliance for Mathematics and Artificial Intelligence in Beijing, an initiative spearheaded by Fields medallist Shing-Tung Yau that will bring together eight leading institutions and aims to build a network of more than 1,000 scientists. The alliance’s stated goals are to develop specialized AI models for mathematics, use the technology to generate conjectures and formalize proofs, and build reasoning systems that are easier to interpret.
Yau, who directs the Yau Mathematical Sciences Centre at Tsinghua University, framed the launch as catch-up more than celebration. “A year ago, I did not think AI was this powerful,” he said, adding that over the past six to twelve months he had watched AI tackle problems that had stumped mathematicians. “The development of AI in mathematics is like a tsunami,” he said, arguing that Chinese mathematicians needed to organize and engage with the technology rather than avoid it.
That urgency is pointed specifically at the United States. Yau said the gap between China and the US was most pronounced in frontier scientific research, particularly in AI systems capable of advancing mathematics and physics. Chinese models handle everyday user tasks well, he said, but “if you ask them to do the work OpenAI has done over the past month, the gap is very large.” He pointed to the scale of mobilization behind US labs: thousands of scholars and heavy industrial investment, with mathematicians helping improve AI systems even as they use those systems to investigate new questions. “The United States and the West have brought together their top mathematical experts to work with AI and explore frontier problems,” he said, calling China’s own efforts fragmented by comparison.
A Pointed Rebuttal To Terence Tao
The timing of the alliance isn’t incidental. It lands in the middle of the loudest fight mathematics has had over AI in years, one largely driven by Fields medalist Terence Tao, who has gone from cautious AI enthusiast to the field’s most prominent skeptic of the pace of adoption.
Tao isn’t anti-AI by any means — he has described using ChatGPT to save himself hours of manual coding and has called current models “ready for primetime” for parts of his research workflow. But he has also warned that AI-generated proofs can “look superficially flawless” while hiding subtle mistakes that a human mathematician would catch instantly, and more recently argued that the industry needs to slow down because “there is no good reason for it to move this quickly.”
Tao’s concerns crystallized into something more formal last month. On September 11, he and 24 other Fields Medal winners — spanning five decades of the prize, from 1978 laureate Pierre Deligne to this year’s winner Yu Deng — published a joint declaration titled “A Severe Misalignment of AI in Mathematics.” The statement argued that the goals of AI companies and the mathematical community are fundamentally misaligned, and that the rush to claim landmark results risks turning mathematics into mass-produced “true/false” statements rather than the kind of understanding that gets built, taught and absorbed by humans over years.
The latest OpenAI release prompted an even sharper rebuke from the Association for Human Mathematics, an international group focused on preserving human authorship in the field, which called the OpenAI release “a demonstration of power” rather than scholarship and urged mathematicians to stop working with the company altogether. That statement followed June’s Leiden Declaration on Artificial Intelligence and Mathematics, a document signed by over 2,600 mathematicians and endorsed by the International Mathematical Union, which warned that AI threatens the integrity of proof, the system of attribution, and the autonomy of mathematical research.
French Fields medalist Hugo Duminil-Copin added one of the more personal entries to the debate — he said that he cried so hard after OpenAI’s release that it woke up his daughter.
Yau Isn’t Buying The Caution
Asked directly about this wave of opposition, Yau was dismissive of the idea that it represented a consensus. “Tao is a very good mathematician, but that does not mean he represents the mathematical community,” he said, adding that the dispute shouldn’t deter mathematicians from working with AI and that what’s actually needed is closer cooperation between leading mathematicians and AI developers.
That framing captures a broader divergence in how the two countries are approaching AI right now. Surveys have repeatedly found Chinese respondents far more optimistic about AI than their American counterparts — Stanford’s AI Index has put Chinese optimism at roughly 84%, more than double the rate in the US — a gap that shows up in how each country’s scientific establishment is reacting to the same technology. Where a large and increasingly organized bloc of American and European mathematicians is publishing joint declarations, boycott calls and essays about what’s being lost, China’s response has been to stand up a 1,000-scientist alliance aimed at closing a capability gap its own leading mathematician says is “very large.”
Whether that scale of coordination can actually close the gap Yau describes is a separate question — one that depends less on enthusiasm and more on the industrial-scale compute and model access that US labs currently have a lead on. But the message from Beijing on Sunday was unambiguous: while Western mathematicians argue over whether to slow down, China’s are choosing to organize and speed up.