OpenAI has formed an external advisory group to give mathematicians a formal channel for assessing its rapidly produced research results. The Advisory Group on Mathematics and Artificial Intelligence will be hosted by the Institute for Advanced Study in Princeton and begins with nine members.
The group can evaluate the significance of new results, coordinate how they are released, offer unsolicited advice and publish its own views. Members will not be paid and control their own membership, features intended to give them some independence from the company.
Its authority has a clear boundary. The advisers cannot slow, redirect or approve OpenAI’s internal mathematics work. The institute emphasized that it has no decision-making power at any AI company, leaving responsibility for OpenAI’s choices with OpenAI. Only one initial member also signed a recent letter from 25 Fields Medalists warning that competition among labs threatens mathematical practice.
The announcement follows OpenAI’s claim that an internal model solved the Navier-Stokes Millennium Prize problem and more than 100 other open problems. Independent review can improve attribution and context around such claims, but the group’s advisory status means its influence will depend on whether the company follows its recommendations.