Summary
An MIT researcher used GPT-5.6 Sol through Codex to select measurement parameters, operate a six-qubit chip, analyse results and refine routine calibration steps. The system worked with little intervention on clear signals but still needed expert guidance when measurements were weak or noisy.
Why it matters
This is a concrete example of an AI agent moving from digital tasks into scientific instrumentation. It suggests that well-defined laboratory workflows may be automated before agents can reliably interpret ambiguous experimental results.
Related coverage
- OpenAI says an internal agent swarm produced a Navier–Stokes disproof
- OpenAI releases ChatGPT Images 2.5 with sharper editing and faster generation
- General Robotics says GRID can engineer a robot’s full deployment lifecycle