A moment to listen

A little more clarity on AI.

Hear one story, understand what changed, and explore the sources behind it.

Evidence as of

EN · 0:51

AI and the Navier-Stokes Controversy

OpenAI claims its AI agents solved a 90-year-old Millennium Prize math problem, but human mathematicians allege their unpublished work may have been used.

Make a show around your interests

No account needed to listen. Your own show starts with the interests you choose.

In this recording

  • OpenAI's claim about an AI-generated Navier–Stokes proof.
  • Why independent verification matters for this Millennium Prize Problem.
  • The dispute over unpublished work and OpenAI's response.

Follow the original reporting

  1. 'Uproar' unleashed after OpenAI says it solved major maths mystery
  2. AI may have solved one of math’s biggest puzzles, raising controversy
  3. OpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hours
  4. OpenAI says it cracked a maths puzzle that has eluded humans for decades in about 88 hours
  5. OpenAI Says It Solved Math's Deepest Problem But Mathematicians Say AI Stole Their Work
  6. On the Navier–Stokes Millennium Prize Problem
  7. OpenAI Says It Solved a $1M Math Problem. A Rival Mathematician Says He Did It First
  8. OpenAI claims to have solved maths problem that stumped humans for decades
  9. OpenAI says it cracked 90-year-old maths problem in 88 hours
  10. Drama swirls around OpenAI’s legendary mathematical milestone
  11. OpenAI has solved the Navier-Stokes Millennium problem using $15m of AI effort
  12. OpenAI Says It Has Cracked One of Math’s ‘Millennium Problems’
Read the transcript

OpenAI claims an unreleased AI model solved the Navier-Stokes existence and smoothness problem. The company says ten thousand AI agents coordinated over eighty-eight hours to generate the proof.

The Navier-Stokes equations deal with fluid mechanics and have lacked a proof for ninety years. It's a Millennium Prize Problem carrying a one million dollar reward. However, the AI's solution hasn't been independently verified.

There is also a dispute. NYU mathematician Tristan Buckmaster and Anthropic's Levent Alpöge say they reached a breakthrough on the same problem in August.

Buckmaster raised concerns that OpenAI might have accessed their unpublished data through its Codex model. OpenAI denied using their work, though it notes de-identified user data can improve models.