LLMs mimic cautious decisions but use different reasoning than humans

Original: Probing Outcome-Level Resemblance and Mechanism-Level Alignment in LLM Risk Decisions: Evidence from the St. Petersburg Game

Source: arxiv.org ↗