The stochastic parrots have apparently proved the Riemann Hypothesis
- Source
- scottaaronson.blog
- Author
- stratos123
- Date

Context
The mathematician Levent Alpöge says that, working with Anthropic's Claude, he has found a proof of the Riemann Hypothesis, and that the method extends to the Generalized Riemann Hypothesis. Scott Aaronson's blog post reports the claim and gives his own reading of it. The Riemann Hypothesis, one of the best-known open problems in mathematics, says every nontrivial zero of the Riemann zeta function has real part 1/2; Alpöge's four-page announcement describes it as asking "for optimal control on the count of prime numbers of a given size." He writes that "Claude and I have also formalized the arguments without assumptions," meaning the proof is written in Lean, a language in which a computer checks every logical step.
What exists so far is that machine-checkable proof, not a paper for mathematicians to read. The code runs to about 17.4 million lines of Lean across more than 175,000 files. Alpöge says the argument is "incredibly long at the moment" and that, rather than release an unreadable write-up, "I decided to only present Lean," while promising to work toward understanding it and writing it properly. A public mirror of the code, which Aaronson links, says its headline theorem uses the Lean math library's own definition of the Riemann Hypothesis rather than one written for this project, and that a text scan found none of the usual markers of skipped steps. The same mirror says "no independent build or review of this code has been published" (mirror).
Aaronson's judgment is that "my bet is that the proof will stand," because "not one of the AI results with Lean certificates has needed to be retracted." He keeps the caveats in view: Lean has had "kernel bugs (which have, in the past, allowed proofs of 0=1)", and he says such weaknesses should be fixed urgently. He also expects that "mathematicians of course won't be satisfied with 17 million lines of Lean" and will want to understand and rebuild the ideas themselves, work he thinks will probably take years. The title's "stochastic parrots" is a dismissive label critics have used for language models, which Aaronson turns into a joke. For now the claim rests on one author's formal proof; the first independent test is other people building that code and confirming what it proves.
Checking sign-in…
Loading comments…


