From Research Goals to Scientific Closure: Evaluating Agents Through Evidence-Supported Physics Research
arXiv:2610.21767v1 Announce Type: new Abstract: Can AI agents organize scientific investigations that lead to well-supported conclusions? To evaluate this capability, we introduce PhysClosureBench, a benchmark of 50 expert-verified, goal-specified research tasks spanning strong-field, attosecond, and neighboring physics areas. Each task specifies the research objective and essential scientific constraints, while leaving the agent to organize the investigation and determine how to generate and v…