Mathematician and AI produce 17-million-line proof of Riemann Hypothesis
A mathematician from Anthropic collaborated with the Claude AI model to produce a 17-million-line Lean proof of the Riemann Hypothesis, which passed independent automated verification. A separate human proof by a Stanford researcher was found to have flaws by AI tools.
The world of mathematics experienced a drama over the past weekend, when two competing claims for solving the Riemann hypothesis were presented almost simultaneously. Levant Alfuga, a mathematician at Anthropic, collaborated with the Claude model and presented a proof in 17 million lines of Lean code, which passed independent automated verification. On the other hand, Alessandro Maria Massolo from Stanford University published a human proof draft of 940 pages, after 12 years of work. Within hours, AI models identified flaws in the human proof. The president of the American Mathematical Society, Ravi Vakil, published an update on the problems found. The path to the million-dollar prize from the Clay Institute requires publication in a peer-reviewed journal and two years of community review.
Mathematician and AI produce 17-million-line proof of Riemann Hypothesis