OpenAI’s Navier-Stokes claim has prompted concern among mathematicians over how powerful AI systems could change research, proof verification and the sharing of unfinished work.
OpenAI said its latest model had produced a solution related to the Navier-Stokes problem, one of seven Millennium Prize Problems announced by the Clay Mathematics Institute in 2000.
The company reportedly deployed about 10,000 AI agents on the task, with the computational effort estimated at $15 million.
Professor Colva Roney-Dougal, head of pure mathematics at the University of St Andrews, said the pace of AI progress had left researchers uncertain about how quickly mathematical research could change.
Professor David Silvester of the University of Manchester said mathematicians could increasingly spend their time checking AI-generated proofs rather than developing them from scratch.
Read: Jacob Coxon Resigns, Says AI 10% Chance of ‘Killing All Humans’
Researchers also raised concerns about academic careers and university teaching. Roney-Dougal said work pursued for months could potentially be overtaken by AI systems, while Silvester questioned the usefulness of unsupervised take-home coursework when students can use AI tools.
Professor Alexander Paseau of the University of Oxford said AI would not remove the intellectual appeal of understanding and appreciating mathematical proofs.
The OpenAI work relied heavily on research by Madrid-based mathematicians Diego Córdoba and Luis Martinez-Zoroa.
Tristan Buckmaster of New York University and Levent Alpöge of Anthropic had also worked on related Navier-Stokes research and questioned whether OpenAI’s model had learned from their unfinished work. OpenAI investigated and denied that it had.
