一作ACL 2026Poster· Natural Language Processing
D ial D efer: A Framework for Detecting and Mitigating LLM Dialogic Deference
Parisa Rabbani · Priyam Sahoo · Ruben Mathew · Aishee Mondal · Harshita Ketharaman · Nimet Beyza Bozdag · Dilek Hakkani-Tür
摘要
LLMs are increasingly used as third-party judges, yet their reliability when evaluating speakers in dialogue remains poorly understood. We show that LLMs judge identical claims differently depending on framing: the same content receives different verdicts when presented as a statement to verify (...