Human scientists are still better than AI ones – for now
A simulator for the process of scientific discovery shows that AI agents still fall short of human scientists and engineers in coming up with hypotheses and carrying out experiments on their own