OpenAI forms nine-member math advisory group (AGMAI) at Princeton's IAS after claiming its model solved 100+ open problems

OpenAI announced an unpaid, independent nine-mathematician Advisory Group on Mathematics and Artificial Intelligence (AGMAI) hosted at the Institute for Advanced Study, after its claim that an internal model resolved more than 100 open math problems; the…

OpenAI has announced the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), a nine-member panel hosted at Princeton's Institute for Advanced Study (IAS). The formation follows criticism of OpenAI's controversial releases of mathematical results, including its claim tying an internal model to a Navier-Stokes Millennium Prize solution. That same model, OpenAI says, has resolved more than 100 additional open problems; The Decoder reports the model solved the long-standing problems after just a month of training. The advisers are unpaid by OpenAI, come from Stanford, Harvard, Oxford, and Cambridge, and include multiple Fields Medalists. They can publicly comment, publish advice OpenAI did not request, and receive early access to OpenAI research. All three reports agree the group is advisory only: OpenAI and IAS say it cannot pace, slow, or redirect the company's internal research. Member Martin Hairer defended the group's independence in a blog post. Only one of the nine initial members, Camillo De Lellis, signed the recent Fields Medalists' letter criticizing AI labs, and while mathematicians welcomed the panel as a first step, some questioned its member selection, actual influence, and representativeness of the math community. OpenAI says its goal is to integrate with the math community and provide guidance on practical applications of mathematical breakthroughs.

• OpenAI announced the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), a nine-member panel hosted at the Institute for Advanced Study (IAS) in Princeton.
• OpenAI says the internal model tied to its Navier-Stokes Millennium Prize solution claim has resolved more than 100 additional open problems; The Decoder reports the model solved the long-standing problems after just a month of training.
• The group was formed after OpenAI's controversial releases of mathematical results and subsequent criticism, including a Fields Medalists' letter criticizing AI labs.
• Members are unpaid by OpenAI, come from Stanford, Harvard, Oxford, and Cambridge, and include multiple Fields Medalists.
• Only one of the nine initial members, Camillo De Lellis, signed the recent Fields Medalists' letter criticizing AI labs.
• The advisers can publicly comment and publish advice OpenAI did not request, with early access to OpenAI research, but cannot pace, slow, or redirect OpenAI's internal research.
• Member Martin Hairer defended the group's independence in a blog post.
• Mathematicians welcomed the group as a first step but questioned member selection, actual influence, and community representativeness.