OpenAI publishes work on 372 outstanding math problems

OpenAI has published solutions or partial solutions to 372 outstanding mathematical problems, according to TechRadar. The problems had challenged mathematicians for years, and the work was published across 722 papers.

TechRadar reports that the solutions are themselves highly complex, making them difficult for mathematicians to understand and verify at scale. The publication has prompted criticism from the Association for Human Mathematics, which Terence Tao shared in a post. The group said mathematicians had not asked for the work and urged them to treat it with due skepticism.

The association also argued that releasing more than 700 files at once demonstrated power rather than scholarship. It said an advisory group on mathematics and artificial intelligence had advised frontier AI companies against testing advanced mathematical problems on internal models. The association accused OpenAI of disregarding scientific research norms and called for science to center human understanding.

OpenAI said it wanted the progress to advance the frontier of human knowledge and enable further progress in mathematics. The publication follows a proposal on the Navier-Stokes equation that OpenAI released a few weeks earlier, generated by a specialized math model that had not been released. TechRadar frames the response as a debate over how AI-generated mathematical work should be published and assessed; the source does not establish whether the published solutions are correct.