The Moment AI Entered the Quantum Experiment Loop

According to AI TIMES, OpenAI revealed on September 8 a case in which GPT-5.6 Sol was linked to Codex and assigned to handle measurement, analysis, and decisions on the next experiment in a quantum-computing experiment. A MIT research group connected the agent to laboratory software on a standard 6-qubit chip to automate measurement and calibration of superconducting qubits. The key point was that AI went beyond helping interpret data and actually controlled equipment through software, while also determining the next conditions based on prior results. In sections with clear signals, it carried out standard procedures with researchers' intervention greatly reduced, but in cases with heavy noise or difficult interpretation, it took more time or required expert judgment. The researchers said the repeated measurement burden was eased, allowing researchers to focus more on analysis, design, and setting direction rather than on monitoring experiments. At the same time, they said they are expanding the agent's use beyond measurement into theory and chip design, while preparing a research environment that assigns multiple tasks in parallel.