Claude and nine-loop physics: what was calculated

In a 25 September 2026 article hosted by Anthropic, physicist Matt von Hippel describes a nine-loop calculation in the N=4 super Yang-Mills model theory. The account and Lance Dixon’s addendum describe Claude-assisted work with human scientific checks. This is a theoretical-method result, not an observation of new particles.

What a loop means here

A loop is not a chatbot conversation round. Readers should consider the mathematical problem and approximation order. Nine loops do not mean nine discoveries. The value rests on a valid calculation rather than a slogan.

Model theory and observation

We distinguish the chosen model, a calculation within it and comparison with measurements. Progress on calculation can matter without establishing an experimental discovery or explaining dark matter.

Verification remains scientific work

Generated code and mathematical expressions are not correct by construction. Assumptions, constraints and sufficiently different checks still matter. Plausible and verified answers may look similar on screen while carrying different evidence.

How to evaluate research agents

A bounded problem with a verification procedure is a useful evaluation setting. Teams should measure the work helped and checking effort required. A single account does not establish a general physics success rate.

Continue reading: antimicrobial research workflows and Claude-assisted enzyme research.