Mathematicians now have a formal channel into OpenAI’s math work, but no lever on its speed. The company announced the Advisory Group on Mathematics and Artificial Intelligence on Monday, hosted at the Institute for Advanced Study in Princeton, New Jersey, and billed as a bridge between the lab and the mathematical community.
Context makes the move conspicuous. A solution to the Navier-Stokes Millennium Prize problem appeared recently, and the same announcement claims the internal model behind it went on to resolve more than 100 other open problems across most of mathematics.
Nine mathematicians joined as founding members, including the institute’s Camillo De Lellis. Of the nine, only he had also signed an open letter this month in which 25 Fields Medal winners accused AI labs of threatening their field by racing to crack famous problems.
Independence comes with limits. Members draw no pay, yet they may volunteer advice, publish their own conclusions, and decide who else joins. Pacing is off the table. OpenAI’s post says the group will not advise on how fast internal math research should move, and the institute’s own release states it holds no decision-making power at any AI company.
