Matt von Hippel writes about how researchers at Anthropic successfully challenged an LLM to solve a complex problem in theoretical particle physics by computing a nine-loop amplitude in N=4 super Yang-Mills. The task involved using the bootstrap method and form-factor approaches through Claude Science, demonstrating that AI can autonomously manage long-running computational tasks with high reliability without constant human oversight.
- The calculation was performed using Fable 5.1 within the Claude Science platform.
- While humans have calculated up to eight loops in this toy model, no one had yet reached nine loops directly.
- The computation cost approximately $100 when using Python and SymPy for the bootstrap method.
- Physicist Lance Dixon validated that the result was correct by checking it against a form-factor approach he has been working on for two years.
- The calculation was performed using Fable 5.1 within the Claude Science platform.
- While humans have calculated up to eight loops in this toy model, no one had yet reached nine loops directly.
- The computation cost approximately $100 when using Python and SymPy for the bootstrap method.
- Physicist Lance Dixon validated that the result was correct by checking it against a form-factor approach he has been working on for two years.
