OpenAI has a pattern of making impressive mathematical breakthroughs and then mishandling their announcement, alienating researchers. Its latest attempt to repair those relations is the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), an independent body of nine elite mathematicians announced on September 21 via a blog post by UCLA professor Terence Tao. The group is meant to advise OpenAI and other frontier AI labs on how to review and communicate emerging results.
AGMAI member Martin Hairer, a mathematics professor at Imperial College London and EPFL, said the group is totally independent: it receives no financial or technical support from OpenAI, and members have no restrictions on what they can say beyond standard confidentiality needed for early access to research. But he acknowledged that OpenAI's own announcement muddied the picture. Many mathematicians learned about the group through OpenAI's louder rollout and assumed it was an OpenAI-appointed body. Hairer said the announcement's phrasing did not help, even though the wording was technically accurate and carefully crafted.
The group's first task is daunting: helping coordinate the release of scores of results from an unreleased OpenAI model, a prospect already stirring dread among researchers. The episode echoes previous controversies, including the credit and scooping disputes that followed OpenAI's solution to a Millennium Prize problem. For AGMAI to succeed, it must first win the trust of mathematicians, and OpenAI's handling of the announcement has not made that any easier.