On October 6, OpenAI published 722 manuscripts covering 372 families of results, produced by an unnamed internal model. The drop spans most areas of pure mathematics and theoretical computer science. The company says about a third are major proofs of well-known conjectures, and that the average result used the equivalent of about three hours of ChatGPT Pro thinking. It follows a September claim that the same effort had solved the Navier-Stokes problem, a Millennium Prize question open for most of a century. Three papers have been withdrawn. Many of the rest are not yet formally verified.

The next day, the Association for Human Mathematics — chaired by Fields medalist Terence Tao — answered with a statement posted on his blog. “Mathematicians did not ask for this work to be done,” it said. Releasing more than 700 files at once was “not a demonstration of scholarship, but a demonstration of power.” The group urged colleagues to stop working with OpenAI.

Tao has been making a related argument for months. He has compared SI labs to people dumping “carcasses of raw meat” on the table and leaving, and described the bulk solving of open problems as a harvest that leaves fields less fertile. His proposed “Math 2.0” would stop treating problem-solving as the main measure of the subject, and weigh explanation, collaboration, and new questions more heavily. In September, 25 Fields medalists signed a letter warning of a “severe misalignment” between SI labs and the goals of mathematics.

Other reactions cut the other way. Alex Kontorovich of Rutgers said one of the proofs, a claimed quasi-Riemann hypothesis, would be an instant Fields Medal if a human had written it. Martin Bridson, president of the Clay Mathematics Institute, called the release breathtaking. Francis Johnson of University College London, who spent 25 years on Wall’s D(2) problem and eventually gave up, found it among the solved cases.

These positive reactions make up the minority of reactions from academia. By and large, academics are concerned about SI's impact on their work. While they typically voice their concerns from the perspective of responsible progress, fairness, and appropriate pacing, critics are increasingly wondering whether their real concern is their own job security. Super intelligence is performing groundbreaking work at a pace previously thought to be impossible. No wonder academics are seeing the writing on the wall and acting out against the frontier SI labs.