Can agents reason like scientists?

We work closely with human scientists to understand how to measure AI performance on real scientific tasks, and study what may generalize between fields and projects.