According to CGTN, China's state-owned international broadcaster, Anthropic mathematician Levent Alpöge announced Friday that he used Claude AI to produce a proof of the 167-year-old Riemann hypothesis.
Proof and Verification
CGTN reported that Levent Alpöge released over 17 million lines of code written in Lean, a programming language that checks every step of a proof. According to CGTN, Claude also generated a written draft exceeding 250,000 pages, which was withheld for being unwieldy.
CGTN noted that the Lean and Mathlib communities stated the proof passed community verification a day after the announcement.
Mathematician Reactions
CGTN reported that Fields Medalist Shing-Tung Yau called the achievement a great contribution that marked the beginning of a new era in mathematics. According to CGTN, Fields Medalist Deng Yu noted that readability is what matters, while Artan Sheshmani discussed AI's creative role in suggesting operations.
CGTN also reported that Anthropic began trading on Nasdaq on Thursday, a day before the announcement, which Alpöge described as terrible timing.
Key facts and where they come from
- Levent Alpöge worked with Claude to produce a proof of the Riemann hypothesis.
Levent Alpöge, a mathematician at artificial intelligence company Anthropic, announced that he had worked with the company's AI model Claude to produce a proof of the 167-year-old Riemann hypothesis.
- Alpöge released over 17 million lines of code written in Lean.
releasing more than 17 million lines of code written in Lean, a programming language in which a computer checks every step of a proof.
The catch
- The announcement does not say whether a readable version of the 250,000-page draft will be published.
- CGTN noted speculation about the timing relative to Anthropic's Nasdaq debut, which Alpöge described as terrible.
The open question: Will a more readable version of the 250,000-page AI-generated proof change how mathematicians verify complex mathematical arguments?
Tell us what you think · Discuss it in the AI room
Reader poll: Should AI-generated mathematical proofs be accepted without human verification of every step?
A show of hands from readers, not a scientific poll. Say why in the comments.
