Yes, Claude can do Nine Loops (www.anthropic.com)

Physicist Matt von Hippel challenged AI companies to tackle complex problems in theoretical physics, specifically within amplitudeology, a subfield focused on scattering amplitudes that predict particle interactions. He proposed that an LLM could solve advanced calculations, which traditionally require significant computational resources, for instance, asking for the computation of N=4 super Yang-Mills amplitudes to nine loops. Anthropic's Claude LLM responded impressively by navigating this computational challenge utilizing a method known as bootstrapping. Remarkably, it achieved this on a modest budget, utilizing about 96 CPUs for a week, making the approach accessible and demonstrating AI's growing capabilities in scientific computation. The significance of this achievement lies in the potential to democratize complex calculations and provide insights into fundamental physics, previously deemed too computationally intensive for conventional methods. This development not only highlights how AI can assist in scientific discovery but also opens the door for further exploration of AI's practical implications in research settings.