Two Fields Medalists Urge Open Models and Source Traceability in AI-Era Mathematics

On Oct. 21, 2026, Yu Deng, a new Fields Medalist and professor at the University of Chicago, and Terence Tao, a Fields Medalist and chairman of SAIR Foundation, discussed the future of mathematics in the age of AI on the SAIR podcast. Both said mathematical models should remain open.

Deng said AI can greatly expand the boundaries of mathematics and research, creating new methods and ideas. Once AI offers an answer to a problem, it often generates several harder new questions that can lead to further exploration. He stressed that openness is a key trait for AI, allowing it to develop healthily and benefit the research community.

Tao said tech companies that use AI for mathematical research are essentially distilling other people's work, since their models draw on published papers and existing literature. That makes it necessary to trace where results come from. He noted that academia has treated scientific literature as a shared resource for centuries, relying on citations to trace the origins of ideas. AI, however, can turn that accumulated literature into answers through a process that is not transparent, potentially breaking the chain of intellectual inheritance. Deng added that being able to see an AI's chain of thought would help in understanding how it reached a result and what references it used, but currently that is not possible, which he called absurd.

Tao also argued that mathematicians should be brought into projects before they begin, rather than being asked to clean up afterward. He said problem-solving in AI-driven work now often emerges as a byproduct of model training, and incentives should shift toward broader understanding and insight, not just solving problems.