OpenAI announced Sept. 21 that it has formed an independent Advisory Group on Mathematics and Artificial Intelligence, hosted at Princeton's Institute for Advanced Study, to help coordinate how the company communicates and releases mathematical results from its models. OpenAI said its AI has resolved more than 100 open mathematical problems across multiple fields, according to the company's announcement and TechCrunch.
The nine-member board includes mathematicians Edward Witten and Timothy Gowers, according to a post on mathematician Terry Tao's blog that carried the group's own announcement as a guest post. The Institute for Advanced Study said in a statement quoted by TechCrunch that it does not have "decision making power at any AI company" and that OpenAI keeps full responsibility for its own decisions. The group "will not be responsible for advising us on how to pace our internal progress on mathematics," TechCrunch reported OpenAI as saying.
The announcement follows weeks of tension after OpenAI said its models had solved the Navier-Stokes Millennium Prize problem, a claim that drew accusations the company was racing to one-up rivals on famous unsolved problems, TechCrunch reported. Twenty-five Fields Medal winners signed an open letter this month warning that AI labs are threatening mathematicians' intellectual work in that race. Only one of the nine advisory group members, Camillo De Lellis, also signed that letter.
An advisory board with no say over pacing and no enforcement power is a fig leaf if OpenAI ignores it and a real check if it doesn't. Which one it turns out to be will show up the next time OpenAI wants to announce a big result before independent mathematicians have checked it.