Saturday, November 21, 2026← Front page
Also this week

Claude physics results

Anthropic says Claude solved two open problems in black-hole physics by running small groups of agents that were copies of itself.12 In the first problem, Anthropic says Claude found that the phase transition in the weakly coupled quantum theory is an abrupt jump as temperature rises, analogous to Hawking and Page's result. In 2003, Ofer Aharony and colleagues had shown the question hinged on a single number appearing three loops in.34 In the second problem, on the modular bootstrap for 3D quantum gravity, Anthropic says Claude proved that the lightest black hole can be no heavier than c/Λ, with Λ approximately 9.1266, and that this ceiling is optimal for this version of the method.56

Earlier numerical work had suggested a ceiling of about c/9.1, while Tom Hartman, Dalimil Mazáč and Leonardo Rastelli had analytically proved only c/8.5, so Claude's exact result closes that gap. inference78 Anthropic adds that the result shows this version of the method cannot on its own push the ceiling down to c/12, so other ingredients will be needed.9

Anthropic says almost every piece of the Hawking-Page-type calculation was done twice by independent agents, with parts also checked against earlier literature.1011 Anthropic says the bootstrap results were formally verified in Lean in September and that a physicist checked every main result.1213 Anthropic also says a physicist who independently used Claude on the Hawking-Page problem later shared a result that matched an earlier agent's answer, reached by a different method.1415

In a separate piece, Timothy B. Lee reports that prominent physicists such as Juan Maldacena are already getting help from AI models, and that Andreas Karch runs AI agents around the clock.1617 Lee speculates that if AI keeps improving, theoretical physics might come to resemble a humanities field, with physicists interpreting knowledge rather than discovering it.1819

What each sentence rests on
  1. The agents solved the two problems in this post. — Anthropic, Claude solves two open problems in the physics of black holes
  2. Claude did it by running a small research group of copies of itself — Anthropic, Claude solves two open problems in the physics of black holes
  3. The result Claude found is an abrupt jump as the temperature rises, exactly analogous to what Hawking and Page had found using the sum over spacetime histories. — Anthropic, Claude solves two open problems in the physics of black holes
  4. In 2003, Aharony and his colleagues realized that whether the phase transition was an abrupt jump, as Hawking and Page predicted, or gradual, hinged on a single number that appeared only three loops in. — Anthropic, Claude solves two open problems in the physics of black holes
  5. It then used it to prove that the lightest black hole in 3D quantum gravity can be no heavier than c/Λ, where Λ is given by an exact formula and is approximately 9.1266. — Anthropic, Claude solves two open problems in the physics of black holes
  6. It also proved that this ceiling is optimal: it’s the lowest this version of the modular bootstrap can ever go. — Anthropic, Claude solves two open problems in the physics of black holes
  7. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  8. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  9. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  10. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  11. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  12. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  13. a physicist checked every main result — Anthropic, Claude solves two open problems in the physics of black holes
  14. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  15. Quote not reprinted: this page already repeats as much of this article as we allow. — Anthropic, Claude solves two open problems in the physics of black holes
  16. Juan Maldacena, another of the field’s eminent figures, acknowledged help from OpenAI models on two papers this year — Timothy B. Lee, I spoke to seven physics experts grappling with the AI revolution
  17. It’s pretty much running 24 hours a day. I always have it do something, — Timothy B. Lee, I spoke to seven physics experts grappling with the AI revolution
  18. In this scenario, theoretical physics might start to look something like a humanities field. — Timothy B. Lee, I spoke to seven physics experts grappling with the AI revolution
  19. But they would not be discovering knowledge so much as interpreting it. — Timothy B. Lee, I spoke to seven physics experts grappling with the AI revolution

Every sentence was checked word for word against quotes from the articles it cites. Each footnote shows its quote, unless the page already quotes as much of that article as we allow; then it links the article instead. A model wrote and arranged the sentences, and a second read checked each against its quotes (written 2026-11-20).

The record: who reported it, checked facts, all quoted views, timeline →