PRINCETON, NEW JERSEY / RankWire.AI / – OpenAI has initiated a partnership with an autonomous group of mathematicians to evaluate and disseminate AI-produced mathematical findings. Hosted at the Institute for Advanced Study in Princeton, the Advisory Group on Mathematics and Artificial Intelligence offers guidance on emerging research, academic standards, and the application of OpenAI tools in scholarly and educational contexts. This panel functions independently from OpenAI, with its members receiving no remuneration from the organization.
The panel is empowered to offer unsolicited advice and publish its findings. It also retains the right to publicly comment on how OpenAI’s work influences the field of mathematics. Membership decisions are within the group’s control, and it may advise other AI firms whose models have a notable impact on mathematics. Importantly, the group has no authority to make decisions on behalf of OpenAI or any other AI enterprise. Furthermore, OpenAI has clarified that the group will not influence the company’s internal pace of mathematical research.
The initial team comprises nine mathematicians: François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten, and Melanie Matchett Wood. They are affiliated with institutions such as the Institute for Advanced Study, Harvard, Stanford, Oxford, Berkeley, Cambridge, EPFL, and ENS-PSL. De Lellis and Witten serve as principal investigators for the Institute for Advanced Study project hosting this initiative.
OpenAI reports rapid advancements in mathematical discovery
OpenAI announced this advisory arrangement amid revealing new achievements related to an internal artificial intelligence model. The company indicated that training on this model began on August 28. Since then, it claims the system has addressed over 100 longstanding open problems spanning most branches of mathematics. These accomplishments have been attributed to outputs generated by the company’s internal model. The advisory group states that its immediate focus is on how OpenAI should coordinate the disclosure of numerous significant results it has reported.
This announcement comes after OpenAI’s publication on September 8 concerning the Navier-Stokes existence and smoothness problem, one of the Millennium Prize Problems. OpenAI detailed an analytical proof and formalization utilizing the Lean proof system. The company claimed that its internal model produced the proof and demonstrated that a smooth fluid could develop a singularity within finite time. However, OpenAI emphasized it does not intend to claim the Millennium Prize for this achievement.
External panel emphasizes standards for validation and dissemination
The advisory group was formed after OpenAI reached out to some of its members about establishing an external advisory council. Instead of the original plan, the mathematicians founded an independent body and extended invitations to additional experts. The group’s mission encompasses interactions between AI organizations, mathematical research, and the broader mathematics community. Its goals include ensuring responsible presentation and release of mathematical results. This effort follows ongoing discussions among mathematicians about how AI-generated research should be evaluated, credited, and shared.
Currently, the group’s work focuses on the mathematical results produced by OpenAI’s internal model. It seeks input from the mathematical community as it crafts guidelines for releasing these findings. The group has committed to publishing its recommendations on its website, with responses from mathematicians kept confidential unless they approve public sharing. Final decisions will rest with the companies receiving the advice, while the group acts as an intermediary connecting mathematicians and the AI industry.