OpenAI’s internal model is claimed to have solved 100-plus open math problems
A user relaying OpenAI’s claim highlights mathematicians’ concern that humans may learn nothing from a model-generated proof.
TLDR
A user says OpenAI claims its internal model has solved 100-plus open math problems, in addition to its Navier-Stokes claim. The user also says OpenAI established an independent math advisory group at IAS / Princeton to review what gets released. Highlighting mathematicians’ concerns about whether AI-generated proofs yield human insight, the user argues that a correct-looking answer still needs human checking.
OpenAI’s internal model is claimed to have solved 100-plus open math problems
A user relaying OpenAI’s claim highlights mathematicians’ concern that humans may learn nothing from a model-generated proof.
TLDR
A user says OpenAI claims its internal model has solved 100-plus open math problems, in addition to its Navier-Stokes claim. The user also says OpenAI established an independent math advisory group at IAS / Princeton to review what gets released. Highlighting mathematicians’ concerns about whether AI-generated proofs yield human insight, the user argues that a correct-looking answer still needs human checking.