Coding Agents and Humans Run Autonomous Synchrotron X-ray Experiments

Researchers used coding-agent infrastructure to run autonomous synchrotron X-ray experiments on nanoparticle superlattices under human-defined objectives that changed over time. Complete human-agent and agent-agent transcripts let them reconstruct decision provenance and measure how work was divided. Across three experiments, 61-67% of messages involved interpretation or judgment, and the share of methodological and scientific contributions rose from 46% in the setup-focused first run to 56-61% later.

Why it matters: The paper offers concrete evidence and an audit method for how agents and humans can share cognitive work in physical scientific instruments, which informs the design of multi-agent systems and agent harnesses.