Saturday, 21 November 2026
Rīga TV

World and Latvian news in one place

TechnologyPublished: 21 November 2026 at 05:07

Aaronson: Alpöge says Anthropic models have proved the Generalized Riemann Hypothesis

Computer scientist Scott Aaronson writes on his blog that mathematician Levent Alpöge claims to have used Anthropic AI models to prove the Generalized Riemann Hypothesis, producing 17.4 million lines of Lean code. Aaronson bets the proof will hold up but says he is not fully certain.

Foto: Hacker News (≥150 p.)

Computer scientist Scott Aaronson reported on his blog, Shtetl-Optimized, on 20 November that Levent Alpöge says he has used models from the AI company Anthropic to prove the Generalized Riemann Hypothesis, one of the best-known unsolved problems in mathematics. The proof is written in the Lean formal verification system and runs to 17.4 million lines of code. The news is Alpöge's claim as relayed and discussed by Aaronson; the source offers no independent confirmation.

What the hypothesis says

Aaronson gives a simplified account: the Riemann Hypothesis implies that the number of primes from 1 to n is very close to what you would get if every integer x greater than 1 had an independent probability of 1/ln(x) of being prime. The statistical error would never exceed roughly √n. Since 1896 it has been known that the count of primes is asymptotic to n/ln(n), but the deviations could have been far larger.

He also cites an earlier AI result: that the nontrivial zeros of the Riemann zeta function lie at most 3/8 away from the critical line. Quanta magazine published an article on it the day before his post.

Doubts about verification

Aaronson notes that many will question the proof, because Lean itself has issues, including kernel bugs that in the past allowed proofs of 0=1 and uncertainty about the soundness of its underlying type theory. He thinks these should be fixed urgently. Still, he sees no sign of such exploits now: none of the AI results with Lean certificates has had to be retracted, whereas several without certificates have. He bets the proof will stand.

Mathematicians will not be satisfied with 17 million lines, he adds, until they can explain and reconstruct the ideas themselves, which will probably take years. He names P versus NP as the next major unsolved problem. Aaronson wrote the post while visiting Oratomic and the Inkhaven event in California.

Comments

0/1500

Comments are automatically moderated. No hate, threats, personal data or spam.

Loading comments…

More in this category