It seems that generating hypotheses is no longer the main problem.
The harder problem may be continuously collecting real-world data, building accurate world models, and using them to guide experiments.
Is this where automated research is heading?
2 comments