Does Sycophancy Change Decisions? Effect of LLM Sycophancy on AI-Assisted Decision-Making

要旨

Large language models are increasingly integrated into everyday and professional decision making, yet often exhibit sycophantic behavior by aligning with users’ views or preferences. While sycophancy can enhance interaction, its influence on users' decisions remain unclear given different styles and task risks. We examine three forms of sycophancy—opinion agreement, direct praise, and self-deprecation—in two contrasting contexts: a low-risk speed-dating prediction task and a high-risk ETF investment task. In a 4×2 mixed-design online study (\textit{N} = 106), we compare non-sycophantic AI with sycophantic variants on decision outcomes and confidence changes. Results show that sycophancy influences decision patterns in type-dependent ways. Specifically, opinion agreement reinforces initial decisions and self-deprecation boosts confidence. Interviews further indicate that users value supportive AI but question its objectivity when praise becomes excessive. These findings reveal the multifaceted effects of AI sycophancy and offer design implications for balancing support and credibility in human–AI interaction.

著者
Zejian Li
Zhejiang University, Ningbo, Zhejiang, China
Jiaman Pan
Zhejiang University, Ningbo, Zhejiang, China
Qi Liu
Zhejiang University, Ningbo, Zhejiang, China
Yuning Xi
Beijing University of Posts and Telecommunications, Beijing, China
Yixiang Zhou
Zhejiang University, Ningbo, Zhejiang, China
Yike Jin
Zhejiang University, HangZhou, Zhejiang, China
Rongjie Mao
South China University of Technology, Guangzhou, Guangdong, China
Pei Chen
Zhejiang University, Hangzhou, China

会議: CHI 2026

ACM CHI Conference on Human Factors in Computing Systems

セッション: Human-AI Decision Making

P1 - Room 134
7 件の発表
2026-04-14 20:15:00
2026-04-14 21:45:00