An Anthropic Mathematician Says He and Claude Proved the Riemann Hypothesis. Within a Day the Checkers Certified All 17 Million Lines, and a Top Mathematician Says 'We Don't Even Know Why the Theorem Is True'
About This Episode
Levent Alpöge of Anthropic announced on Friday, November 20 that he and Claude had found a proof of the Riemann Hypothesis, the most famous open problem in mathematics, and released it as roughly 17 million lines of machine-checkable Lean code instead of a paper. By Saturday the Lean and Mathlib communities had certified both the statement and the proof as correct, while the Clay Institute, Kevin Buzzard and Alpöge himself said human understanding of the argument has not caught up. An all-human rival draft posted from Stanford hours later is the one reviewers are finding problems in.
Our Take
The most famous problem in math now has an answer a machine can certify and no person can yet explain: trust moved from the mathematician to the checker in a single day, and understanding is the part still owed.
Continue Reading on Unscarcity
The Verifiability Premium: When Proof Replaces Credentials
Direct match: the article's two-cost model (generating an answer versus checking it) and its list of what a Lean certificate cannot tell you is exactly the split this story lands in, where the machine's proof is certified in a day and the human draft is the one in doubt.
Amazon's AI Penalized Women. Caesar Became Dictator.
Its reading of Law 2, Truth Must Be Seen, sets the bar at a human being able to trace and understand the reasoning, not merely at publication, which is the gap between 17 million open lines and nobody knowing why the theorem is true.