PRINCETON, NEW JERSEY / RankWire.AI / – OpenAI has begun working with an independent group of mathematicians on the review and release of AI-generated mathematical results. The Advisory Group on Mathematics and Artificial Intelligence is hosted at the Institute for Advanced Study in Princeton. It will advise on emerging findings, academic standards and the use of OpenAI tools in research and learning. The group operates independently from OpenAI. Its members receive no payment from the company.
The group can provide advice that OpenAI has not requested and can publish its recommendations. It can also comment publicly on the company’s impact on mathematics. Its members control the group’s membership and may advise other AI companies whose models significantly affect mathematics. The group has no decision-making authority at OpenAI or any other AI company. OpenAI has also said the group will not advise the company on the pace of its internal mathematics research.
Nine mathematicians make up the initial group. They are François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer and Nikhil Srivastava. Ulrike Tillmann, Ravi Vakil, Edward Witten and Melanie Matchett Wood complete the membership. Their institutions include 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 that hosts the initiative.
OpenAI reports rapid progress in mathematical research
OpenAI announced the advisory arrangement as it disclosed new claims about an internal artificial intelligence model. The company said it began training the model on August 28. OpenAI says the system has since resolved more than 100 long-standing open problems across most areas of mathematics. The company has described those results as work produced by its internal model. The advisory group says its immediate task concerns how OpenAI should coordinate the release of a large number of significant reported results.
The disclosure follows OpenAI’s September 8 publication of work on the Navier-Stokes existence and smoothness problem. That problem is one of the Millennium Prize Problems in mathematics. OpenAI published an analytical proof and a formalization using the Lean proof system. The company said its internal model produced the proof and showed that a smooth fluid can develop a singularity in finite time. OpenAI also said it does not intend to claim the Millennium Prize for the result.
Independent group focuses on review and release standards
The advisory group formed after OpenAI approached some of its members about creating an external advisory board. The mathematicians instead established an independent group and invited additional members to join. Its stated purpose covers interactions between AI companies, mathematical research and the wider mathematics community. The group plans to address the responsible presentation and release of mathematical results. Its work also follows wider discussion among mathematicians about how AI-generated research should be evaluated, attributed and communicated.
The group’s current work centers on the mathematical results that OpenAI says its internal model has produced. It is seeking input from mathematicians as it develops recommendations on how those findings should be released. The group says it will publish its recommendations to AI companies on its website. Responses from mathematicians will not become public without their approval. Final decisions remain with the companies receiving the advice, while the independent group serves as a channel between mathematicians and the artificial intelligence industry.