OpenAI forms independent mathematics advisory group with nine researchers

OpenAI announced a nine-member advisory group of mathematicians to guide how the company assesses, communicates and releases AI-generated mathematical results. The structure gives researchers a formal channel to publish advice that OpenAI did not request, but no authority over company decisions.

Members will not be paid by OpenAI. They may comment publicly on the company’s effect on mathematics and publish their advice. The group says it will post recommendations on its own website and control its membership.

The advisory group sits outside OpenAI’s formal board governance. Its role is strictly advisory: it has no decision-making power at OpenAI or any other AI company. Responsibility for company decisions stays with the companies receiving its advice.

The mandate has another explicit boundary. OpenAI says the group will not advise it on how quickly to advance its internal mathematics work. Instead, its remit covers assessing the significance of emerging results, coordinating how results are disseminated, developing standards for mathematical research, and considering how AI tools can support research and learning.

For now, the group is advising OpenAI on coordinating the release of mathematical results that the company says came from an internal model. The number and significance of those results have not been independently established.

According to the group, OpenAI initially approached some of the researchers about forming an external advisory board. The researchers and the company instead agreed to establish an independent group and invite the remaining members. The Institute for Advanced Study lists the initiative among projects hosted by its Jonathan M. Nelson Center for Collaborative Research.

The initial members are François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten and Melanie Matchett Wood.