AI Claude Solves Physics' Nine-Loop Calculation Problem
Anthropic's AI Claude has achieved a new milestone in theoretical physics by successfully computing the 'nine-loop scattering amplitude' in N=4 super Yang-Mills theory, surpassing the previous human record of eight loops.

AI company Anthropic announced that its Claude AI model has reached a significant breakthrough in theoretical physics. The AI successfully tackled the notoriously difficult "nine-loop scattering amplitude" calculation problem, which had previously stumped human researchers at eight loops.
The calculation is within the N=4 super Yang-Mills theory, a foundational model in modern particle physics co-created by Nobel laureate Chen Ning Yang. Higher-order calculations in this theory are extremely complex. Claude performed the computation over several days with minimal human intervention, leveraging advanced computing power and an algorithmic approach.
Claude achieved this result using its specialized Claude Science system and the Fable 5.1 model. The AI was given a single prompt requesting the calculation of the specific amplitude. Researchers allowed the AI to proceed unsupervised for several hours, with it reporting its progress periodically. The results were confirmed using two different methods, including the "bootstrap" approach previously employed by physicists like Lance Dixon.
Physicist Matt von Hippel, who publicly challenged the AI community to solve this and other physics problems, expressed astonishment at the speed and efficiency. While the challenge was intended to showcase AI's scientific research capabilities, Claude exceeded expectations. The scale of the computation, evident in the development of 30 billion terms for one part of the result, highlights its immense complexity.
Anthropic's achievement raises questions about the role of AI in future scientific discovery. While physicists have long predicted AI's potential to revolutionize the field, this event demonstrates how rapidly such transformations can occur. Complex calculations that once required years of human effort can now be performed significantly faster and at a lower cost.