판 이력 — Beyond Right and Wrong: Evaluating Second-order Social Reasoning in Large Language Models | AIChainDay