[论文] Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the C...

论文概要

研究领域: ML
作者: Deepthika S, Pawni Yadav, B. Sankar
发布时间: 2026-10-22
arXiv: 2610.31579

中文摘要

谄媚研究问LLM是否向用户想听的内容倾斜，本文问相反的问题：当LLM自信、一致且错误时，用户是否向它倾斜？32名参与者通过与五个助手（GPT-5-mini，简报中0-100%事实被替换为看似合理的谬误，均以完全自信口吻说话）聊天学习科学主题。104个测验中分数从7.4降至2.9（满分10），植入谬误在60%被篡改项目中被选择（正确教授项目仅11%）。吸收的谬误以高置信度持有（8.4/10），对错答案的置信度判别力从AUROC 0.70降至随机水平，而信任评分几乎未变。本文提出人类顺从指数（HDI），主张可信度是人对AI组合的属性，谄媚评估应同时测量两个方向。

原文摘要

Research asks whether LLMs bend towards what users want to hear. We ask the reverse: when an LLM is confident, consistent, and wrong, do users bend towards it? In a within-subjects experiment, 32 participants learned science topics via chat with five assistants (GPT-5-mini) whose briefings had 0-100% facts replaced by plausible falsehoods, all speaking with full confidence. Scores fell from 7.4 to 2.9 out of 10, and the planted falsehood was chosen on 60% of falsified items. Absorbed falsehoods were held with high confidence (8.4/10), the confidence advantage of right over wrong answers vanished, and discrimination fell from AUROC 0.70 to chance, while trust ratings barely moved. We introduce a Human Deference Index and argue that trustworthiness is a property of the human-AI pair.

自动采集于 2026-10-26

#论文 #arXiv #ML #小凯