Former OpenAI Post-Training VP Responds to Terence Tao: Don't Frame AI's Future by Today's Limits

Summary

A statement signed by 25 Fields medalists, posted by Terence Tao on September 11 under the title 《人工智能在数学中的严重错位》, argues that AI companies' focus on solving famous mathematical problems seriously diverges from the community's goals of conceptual understanding, method development, and knowledge transmission. Former OpenAI VP of post-training Liam Fedus responds that today's AI limits should not frame future possibilities: with broader training goals, AI could offer insights and build conceptual frameworks mathematicians can extend — a view he links to Noam Brown's judgment that AlphaGo raised human players' levels. The article recounts that in May an internal OpenAI model rebutted the nearly 80-year-old Erdős unit distance conjecture, and in early September about 10,000 autonomous AI agents found a proof of singularities in the three-dimensional Navier–Stokes equations in about 88 hours, formalized with Lean, pending confirmation by the mathematical community. Disputes over authorship and priority followed.