OpenAI turns to mathematicians as questions mount over AI results
A nine-member advisory group hosted by the Institute for Advanced Study will advise on reviewing and releasing mathematical findings attributed to OpenAI’s model. Mathematicians have questioned its selection and influence.
OpenAI has established a nine-member Advisory Group on Mathematics and Artificial Intelligence to advise on how AI companies engage with mathematical research. Hosted by the Institute for Advanced Study, the group’s remit includes evaluating and communicating emerging results, academic standards, and AI’s effects on research.
The group says its first priority is advising OpenAI on coordinating the release of a large number of mathematical results that the company attributes to an internal model. OpenAI has claimed the unreleased model resolved more than 100 long-standing open problems, including the Navier–Stokes Millennium Prize problem. The material provided contains no independent assessment of those claims.
OpenAI says the group can offer unsolicited advice, comment publicly and publish recommendations. Members are unpaid and may receive early access to the company’s research, according to the announcement. Member Martin Hairer said the group formed after OpenAI approached some mathematicians and is gathering views from colleagues and through a public feedback form.
Researchers quoted by The Verge welcomed engagement but raised questions about how members were selected, whether a small group of prominent academics can represent the wider field, and how much influence it will have. The group itself acknowledges it is not a representative sample of mathematicians.
The panel’s work follows disputes over how AI-related mathematical results have been attributed and communicated. Its recommendations and OpenAI’s response will help determine whether the group can address those concerns as the company prepares to release further results.
