Auditable Human-Agent Teams for Autonomous X-ray Experimentation
Summary
- Researchers demonstrated autonomous synchrotron X-ray experiments on nanoparticle superlattices using coding-agent infrastructure.
- The system operated under evolving, human-defined objectives while maintaining complete human-agent and agent-agent interaction transcripts.
- Analysis of three experiments showed that 61-67% of messages involved interpretation or judgment, with methodological contributions increasing over time.
- Decision provenance revealed that 59-68% of error corrections involved multiple participants, and about half of all decisions incorporated external input.
Why it matters
This study demonstrates how auditable multi-agent human-AI teams can successfully conduct complex, physical scientific experiments, offering insights for future agent harness design and autonomous laboratory infrastructure.