OpenAI introduces mathematics advisory panel amid breakthrough in ai-driven problem solving

OpenAI announced on September 22 the formation of an independent advisory group focused on mathematics and artificial intelligence, based at the Institute for Advanced Study in Princeton. The group comprises nine mathematicians who will operate autonomously from the company. The announcement also revealed that a mysterious model, trained since August 28, has solved over 100 long-standing mathematical problems, including the Navier-Stokes Millennium Prize Problem. OpenAI emphasized that the model's achievements have surprised internal mathematicians, prompting discussions on how to share the findings with the academic community. The company highlighted the potential of AI in advancing basic science and the need for cautious deployment of such capabilities. Previously, mathematicians like Terence Tao, Deng Yu, Peter Scholze, and Camillo De Lellis expressed concerns about the implications of using mathematical problem-solving as an AI benchmark. The advisory group, including Fields Medalists Timothy Gowers, Martin Hairer, and Edward Witten, will guide OpenAI on reviewing and disseminating results, assessing their significance, and supporting mathematical research. The group will not influence the pace of internal research but will provide recommendations on academic standards and tool development for mathematics.