The 2 papers had been posted on 7 November on the Hexagon preprint server, with the creator given as Claude (Anthropic). Each are labelled as primarily AI-generated textual content, with just some components marked as understood by a human.
A research post published by Anthropic on 17 November describes the work as two results from Claude Fable 5.2. In August, Anthropic researcher Hongbin Chen requested Claude to record open issues in theoretical physics, then set brokers to work on 22 candidates for about six weeks, the put up says. The corporate says the brokers solved two of them, largely on their very own.
The 42-page first paper takes up a calculation that physicists set out in 2003 however didn’t end. It asks whether or not a sudden change tied to the formation of black holes in a theoretical universe — the Hawking-Web page transition — nonetheless happens in a quantum description that applies when spacetime now not has a particular form. Anthropic says the brokers discovered that the change is abrupt. The paper itself says it reveals solely that this begins in that restricted regime, not that it continues into the unique gravity setting. A physicist who later used Claude on the identical downside reached the identical reply by a special technique, the put up says.
The 131-page second paper considerations a less complicated mannequin universe with two dimensions of house. Since 2009, the only type of a way referred to as the modular bootstrap has been used to set a ceiling on how heavy the lightest black gap in that universe could possibly be, ruling out theories that break it. Anthropic says Claude discovered the precise restrict for that easiest model — about c divided by 9.1266, the place c measures the scale of the speculation — and confirmed that this model can’t by itself attain c divided by 12, the mark at which black holes first seem. The put up says the outcomes had been formally checked in September in Lean, a proof assistant, and {that a} physicist checked each most important outcome. Neither paper has been peer reviewed. Anthropic says neither outcome settles the query of quantum gravity, or comes shut.

