OpenAI Prompts Independent Math Advisory Panel

On September 21st, a group of prominent researchers announced the formation of the Advisory Group on Mathematics and Artificial Intelligence (AGMAI). The independent body of nine elite mathematicians, which includes Fields Medalist Terence Tao of UCLA and Martin Hairer of Imperial College London, was formed after OpenAI approached several members about establishing an advisory board. While OpenAI initiated the discussions, AGMAI operates independently without financial or technical support from the startup. The group aims to advise frontier AI labs on reviewing and communicating emerging mathematical results.

The urgency behind the group's formation stems from OpenAI's aggressive push into mathematics. The company claims its unreleased internal model has resolved more than 100 long-standing open problems across various mathematical fields. Previously, OpenAI drew criticism for its handling of a purported solution to the Navier-Stokes Millennium Prize problem and its early August announcement of "Ten advances in mathematics and theoretical computer science." Academics accused the company of sloppy scholarship, with Hairer describing the practice of quietly altering manuscripts as "really bad and sloppy scholarship."

For active practitioners, OpenAI's secretive stockpile of solutions has created deep professional anxiety. Colva Roney-Dougal, a mathematics professor at the University of St Andrews, expressed concern that her ongoing work "suddenly becomes irrelevant" as new findings emerge. Álvaro Lozano-Robledo, a mathematics professor at the University of Connecticut, argued that AI companies still fundamentally need human expertise to evaluate and celebrate these solutions. He criticized OpenAI's grandstanding, noting that claiming "substantial progress" on a Millennium Prize problem is not how academic mathematics is conducted.