Skip to content
AI.info

The Pulse

Human-written Riemann proof draft draws AI-assisted challenges

Ravi Vakil posted Alessandro Maria Masullo's proposed proof of the generalized Riemann hypothesis, saying it was developed without AI. Hours later, he said mathematicians had reported problems identified with the help of AI models.

Human-written Riemann proof draft draws AI-assisted challenges

A human-written draft claiming to prove the generalized Riemann hypothesis has drawn reports of problems identified with the help of AI models. In a November 20 post presenting Alessandro Maria Masullo's proposed proof, Stanford professor Ravi Vakil said the work had been developed without AI and asked mathematicians to check it.

Source math216.wordpress.com

About eight hours later, Vakil updated the post to say several people had contacted him about problems in the draft found with the assistance of different AI models. He said the issues would have to be resolved to the mathematical community's satisfaction before the proof could be accepted, and made no claim about whether its strategy was correct.

A draft brought forward

The draft had already received months of attention from Brian Conrad, whose extensive feedback was being incorporated, according to Vakil. Conrad's assessment had so far been positive, Vakil said, and the earlier hope was for him to work with Masullo for a year to polish the writing before an announcement and broader review.

Instead, Vakil posted the current draft while attending an American Mathematical Society meeting. He called for people to check and formalize the argument and to fill any small gaps that might arise.

Refine, which said it had audited the manuscript, The Music of the Primes, separately claimed that the argument had fundamental problems in several parts of its strategy. Its assessment described the manuscript as claiming an unconditional proof of the generalized hypothesis.

A second Riemann claim

The human-written draft appeared on the same day that TechCrunch reported a separate claim by Anthropic mathematician Levent Alpöge, who said he had worked with the company's Claude model on the Riemann hypothesis. His claimed solution was expressed in Lean, an open-source language used to represent mathematical ideas as code and formally verify proofs.

TechCrunch reported that several participants in a mathematicians' discussion, including MIT mathematician Andrew Sutherland, said they had successfully compiled Alpöge's solution. The outlet reported that Alpöge appeared to feel guilty about releasing the proof without a truly human-intelligible explanation.

Years of work

TechCrunch reported that Masullo described his draft as the result of 12 years of work. In an introduction to the proof, he addressed mathematicians discouraged by the pace and tone of the current climate, writing: “we are still here. We are very much alive.”

The two drafts make different claims: Masullo's concerns the generalized Riemann hypothesis, while Alpöge's concerns the Riemann hypothesis. TechCrunch described the latter as a conjecture about a hidden structure underlying the appearance of prime numbers.

Sources

  1. Original source Math 216: Foundations of Algebraic Geometry math216.wordpress.com

Other pages read

  1. TechCrunch techcrunch.com
  2. Refine x.com

Related stories

Explore

More articles