Back to feed
arXiv cs.AI·

Human Adults and LLMs as Scientists: Who Benefits from Active Exploration?

Signal
72
Hype
15
In three linesComparative study showing active exploration substantially improves conjunctive causal reasoning in human adults, contrary to passive observation paradigms. State-of-the-art LLMs approach human-level inference accuracy but exhibit less efficient exploration strategies and similar conjunctive-disjunctive performance gaps.
Read source
Your take?
ReasoningEvals

Summary generated by Claude — human-verified