一作ACL 2026Findings· Findings of ACL
Evaluating Implicit Biases in LLM Reasoning through Logic Grid Puzzles
Fatima Jahara · Mark Dredze · Sharon Levy
摘要
While recent safety guardrails effectively suppress overtly biased outputs, subtler forms of social bias emerge during complex logical reasoning tasks that evade current evaluation benchmarks. To fill this gap, we introduce a new evaluation framework, PRIME ( \textbf{P} uzzle \textbf{R} easoning...