Human insight remains essential as AI tackles mathematical proofs

OpenAI has announced that an AI model solved the Navier-Stokes problem, a longstanding mathematical challenge involving equations that describe fluid movement. On October 6 2026, the company also presented proposed solutions to more than 300 longstanding unsolved problems using a new internal model. These claimed advances could reshape mathematical research, but major AI-generated results still require human expertise to check their reasoning.

Verification is only part of the work. Mathematicians also seek to understand why a result holds and what its proof reveals. Terence Tao, a prominent supporter of AI in mathematics, calls this process “proof digestion”: reorganising arguments, identifying more natural concepts, connecting findings to existing mathematics and extracting methods for other problems. That process can itself generate new mathematics.

Even a verified breakthrough leaves further questions. Important variants of the Navier-Stokes problem remain beyond the methods of OpenAI’s proposed proof, while alternative proofs of established results can reveal fresh connections and approaches. As AI becomes more effective at generating proofs, human work on explanation, interpretation and developing new research questions may become increasingly important. Mathematical progress depends not just on obtaining correct answers, but on making productive use of them.