Daily AI news, sourcedOfficial news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- Fields medalist Richard Borcherds says human math research could be redundant in a few years after OpenAI's Navier-Stokes result, per The Daily Californian
UC Berkeley mathematician Richard Borcherds, who won the 1998 Fields Medal, told The Daily Californian he had not expected AI to reach the level of OpenAI's announced Navier-Stokes result for about 10 years. "It's fairly obvious that if things go like this, in a few years, humans doing math research will be redundant," he said, adding that it is hard to know what to tell graduate students. The paper notes that OpenAI says an internal model running about 10,000 agents produced the result, and that OpenAI researcher Noam Brown said it cost millions of dollars, per The Daily Californian.
Reported by The Daily CalifornianPublished newsSource: The Daily CalifornianPart of OpenAI math releases - Ben Recht says OpenAI's claimed faster Fourier transform is a galactic algorithm with no practical speedup, per arg min
Berkeley's Ben Recht examined the faster-FFT result in OpenAI's recent math paper release. He argues the hidden constant is around 10^100, so Cooley-Tukey stays faster for any realistic input, and that the proof assumes exact complex arithmetic on numbers that grow superpolynomially, a model no real computer matches. He calls the result an artifact of loose formalization rather than a breakthrough and makes a similar critique of the O(n^2.25) matrix multiplication result in the same release.
- Researchers say OpenAI’s Navier–Stokes Lean formalization does not match its natural-language proof, per New Scientist
Cambridge-linked mathematicians argue OpenAI’s Lean formalization of its Navier–Stokes blow-up result is not a faithful translation of the natural-language PDF proof, so Lean verification does not certify the prose argument, per New Scientist. They highlight Lemma 8.6, where the NL statement requires a bound of m+4 while the Lean estimate uses the weaker m+5, and stress they are not claiming either proof is false. OpenAI told New Scientist it knows of the mismatch, says that does not invalidate either proof, and will fix NL errors as found while continuing to formalize a larger batch of papers.
- Association for Human Mathematics urges boycott of OpenAI after AI math manuscript dump, per The Decoder
The Association for Human Mathematics called on mathematicians to stop working with OpenAI after its October 6 release of 700+ AI-generated manuscripts, calling the dump a show of power that ignored AGMAI advice against testing advanced math on internal models, per The Decoder. Terence Tao shared the AHM statement as a guest post and argued Math 2.0 should value explanation over raw problem-solving; Scott Aaronson described a “Mathocalypse” as experts struggle to read opaque drafts. AGMAI took a milder line, treating the release as a starting point while warning research cannot become cleanup of lab outputs.