AI did not just help with a famous mathematical proof. It helped make the proof machine checkable. Anthropic says @claudeai worked largely autonomously for 11 days to produce the first complete computer checked formalization of Fermat’s Last Theorem in Lean. 13 million lines of formal proof. 29,500 intermediate theorems used in the final result. That distinction matters. Andrew Wiles already proved Fermat’s Last Theorem decades ago. The breakthrough here is turning an extraordinarily complex body of human mathematics into something a computer can verify end to end. And I think this points to one of the most important roles AI may play in science. Not replacing the scientist. Reducing the verification bottleneck. As AI generates more mathematics, code and scientific reasoning, producing answers will become cheaper. Proving that those answers are actually correct may become more valuable. The next frontier of AI might not only be intelligence. It might be verifiable intelligence.