OpenAI has said that its internal AI models have solved three unresolved problems in mathematics, computer science, and physics. Among the experts invited to verify the proofs is a Nobel laureate; they are assessing whether the proposed proofs withstand scientific scrutiny.
One of the problems is related to a Millennium Prize Problem; for solving each such problem, a prize of 1 million dollars is awarded. According to The Decoder, if the computer science proof is confirmed on its own, hundreds of scientific papers could lose their footing.
The significance of the news still depends on external validation: the available package includes only a published The Decoder brief description, not the texts of the proofs or the experts’ conclusions. Therefore, OpenAI’s statements should be treated as a report awaiting independent assessment.
editorial commentary
Why it matters
A likely consequence is increased scrutiny of independent verification of AI results in fundamental science. The next observable signal will be the published conclusions of the invited experts. Substantial uncertainty remains: the initial information is based on a single brief The Decoder post and does not contain the proofs themselves.