Anthropic says Claude solved two open black hole physics problems
Anthropic says Claude solved two open black hole physics problems, including a more than 20-year-old calculation on whether a black hole phase transition remains abrupt without definite spacetime. The company said the model used copies of itself to perform and check the calculation, and posted preprints. The result gives physicists a concrete test case for using AI in theoretical research.
Anthropic said Claude solved two open problems in theoretical physics involving black holes and the search for quantum gravity. The first completed a calculation proposed by physicists in 2003 about whether a black hole formation transition remains abrupt when spacetime becomes blurry. The company said the answer was yes. The second found the exact limit of the simplest version of a technique used to test candidate theories of quantum gravity.
Anthropic said the model worked as a small research group of copies of itself, solving nearly every piece of the calculation independently and checking its mathematics against known results. The company posted the findings as preprints, not as a peer-reviewed journal article. The source provides the claimed results and method, but not independent confirmation from external physicists. The key evidence is the completed calculation itself, described as exact and checked against established cases.
The physics setting is holographic duality, a tool that turns hard gravity questions into quantum questions on a lower-dimensional boundary. Stephen Hawking and Don Page showed in 1983 that heating a model universe can cause an abrupt switch from particle gas to black hole, but their calculation assumed large, smooth spacetime. Physicists later found a related transition in the quantum description, yet could not tell whether it was abrupt or smooth. Claude's claimed result connects the two by showing abruptness survives without definite spacetime.
For physicists, the result addresses a long-standing question about whether black hole formation is a feature only of classical spacetime or is encoded in the underlying quantum theory. For AI developers, the method shows a model can be used to decompose, execute, and cross-check a long research calculation rather than only produce prose. The stakes are concrete: the claimed exact result in a simplified model gives theorists a new benchmark for machine-assisted scientific reasoning.
Theoretical physicists get a claimed exact result about black hole formation in simplified quantum-gravity models. AI developers get a concrete example of using a model to run and cross-check a long scientific calculation, not just write text.
Sources (1)
- [1]Original source(https://www.anthropic.com/research/claude-black-hole-physics)