OpenAI forms elite mathematicians' panel after backlash over research disclosures
Following months of criticism over how it announced mathematical breakthroughs, OpenAI has created an independent nine-member panel of mathematicians to advise it. Researchers call it a positive step but question its actual influence.
OpenAI announced on Monday the creation of an independent advisory group of mathematicians tasked with guiding the company and other AI developers on how they interact with mathematical research. The group, called the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), will be hosted by Princeton's Institute for Advanced Study, a revered research center once home to figures such as Albert Einstein and John von Neumann.
The nine-member panel includes prominent mathematicians from institutions such as Stanford, Harvard, Oxford, and Cambridge, several of whom hold Fields Medals or MacArthur "genius grants." OpenAI said the group will operate independently, with freedom to offer unsolicited advice, comment publicly on the company's impact on mathematics, and publish its own recommendations. Members will not be paid by OpenAI and the group can adjust its own membership.
Questions over origins and influence
Mathematicians who spoke with The Verge described the initiative as a good first step but said they remain uncertain how much real influence the group will have, whether OpenAI will actually act on its advice, and whether nine people can adequately represent such a broad field. Group member Martin Hairer wrote in a blog post that the organization is genuinely independent of OpenAI and other AI labs, that members have signed no restrictive agreements, and that none receive financial compensation.
Hairer acknowledged that AI companies have in recent months released high-profile mathematical results in ways that fall well short of acceptable academic standards. He said the group's first priority is advising OpenAI on how to coordinate the release of a large number of significant results the company says were produced by its internal model.
Background: disputes over credit and scale of results
The new effort follows months of criticism of OpenAI's conduct, including accusations that it scooped researchers and failed to credit human contributions underlying its models' work. Alongside announcing the panel, OpenAI disclosed that its unreleased model — the same one that resolved the Navier-Stokes Millennium Prize problem — has now resolved more than 100 long-standing open problems across most areas of mathematics.
