OpenAI sets up a maths advisory group after claiming 100+ solved problems

OpenAI announced an Advisory Group on Mathematics and Artificial Intelligence, hosted at Princeton's Institute for Advanced Study, with nine mathematicians on it. The announcement came with a claim: an internal model has now resolved more than 100 open problems across several fields.

The group exists because of a backlash. After OpenAI said on September 8 that a model had produced a counterexample settling the Navier-Stokes Millennium Prize problem, twenty-five Fields medallists signed an open letter arguing that AI labs are threatening mathematicians' intellectual work as they compete to out-announce each other. Only one of the nine advisers, Camillo De Lellis, signed that letter.

The group has no authority and no pay. OpenAI stated that it “will not be responsible for advising us on how to pace our internal progress on mathematics” — the same word, pacing, that the labs are using everywhere else this month, and the same pattern: an outside body invited to observe rather than to decide.

• Reported OpenAI says an internal model has resolved more than 100 open problems across several fields. TechCrunch
• Reported Twenty-five Fields medallists signed an open letter saying AI labs are threatening mathematicians’ intellectual work. TechCrunch
• Confirmed OpenAI: the group “will not be responsible for advising us on how to pace our internal progress on mathematics.” TechCrunch