Claude-assisted Riemann proof claim faces expert scrutiny
Anthropic mathematician Levent Alpoge claims to have proved the generalised Riemann hypothesis with help from Claude. The claim remains unverified, and no human has read through the entire argument. If confirmed, it would resolve a conjecture extending the Riemann hypothesis, which concerns the hidden order underlying the distribution of prime numbers.
Instead of a conventional mathematical paper, Alpoge released his argument in Lean, a programming language and proof assistant that checks reasoning step by step. Imperial College London mathematician Kevin Buzzard said the submission contains 17 million lines of Lean code. Researchers worldwide are translating portions into ordinary mathematical language to determine what it establishes and whether it introduces new ideas or combines existing techniques.
A confirmed proof would have consequences across mathematics, where many results depend on the hypothesis being true. It could also establish the reliability of algorithms whose correctness depends on the generalised version, including algorithms relevant to internet security. Experts expect a clearer understanding of the submission within weeks.
The claim highlights AI’s growing role in mathematical research. Queen Mary University of London mathematician Abhishek Saha said directing AI, choosing research questions and explaining results could become increasingly important skills. He stressed that Alpoge deserves credit if the proof holds up, while Buzzard described AI as a powerful tool rather than a replacement for human mathematicians.