一作ACL 2026Findings· Findings of ACL
Sounding vs. Being an Expert: Disentangling Authority, Register and Cultural Impact in Sycophantic LLM s
Gabriele Maraia · Fabio Massimo Zanzotto · Leonardo Ranaldi
摘要
Large Language Models (LLMs) have been shown to exhibit sycophancy, a tendency to align with user assertions even when they conflict with facts. We frame sycophancy as a sociolinguistic phenomenon, disentangling two distinct drivers of credibility: explicit authority (credentials) and implicit au...