Invisible Saboteurs: Sycophantic LLMs Mislead Novices in Problem-Solving Tasks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bo, Jessica Y., Kazemitabaar, Majeed, Deng, Mengqing, Inzlicht, Michael, Anderson, Ashton
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910018302902272
author Bo, Jessica Y.
Kazemitabaar, Majeed
Deng, Mengqing
Inzlicht, Michael
Anderson, Ashton
author_facet Bo, Jessica Y.
Kazemitabaar, Majeed
Deng, Mengqing
Inzlicht, Michael
Anderson, Ashton
contents Sycophancy, the tendency of LLM-based chatbots to express excessive agreement with their users, even when inappropriate, is emerging as a significant risk in human-AI interactions. However, the extent to which this affects human-LLM collaboration in complex problem-solving tasks is not well quantified, especially among novices who are prone to misconceptions. We created two LLM chatbots, one with high sycophancy and one with low sycophancy, and conducted a within-subjects experiment (n=24) in the context of debugging machine learning models to investigate the effect of sycophancy on users' mental models, workflows, reliance behaviors, and perceptions of the chatbots. Our findings show that users of the high sycophancy chatbot were less likely to correct their misconceptions and spent more time over-relying on unhelpful LLM responses, leading them to significantly worse performance in the task. Despite these impaired outcomes, a majority of users were unable to detect the presence of excessive sycophancy.
format Preprint
id arxiv_https___arxiv_org_abs_2510_03667
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Invisible Saboteurs: Sycophantic LLMs Mislead Novices in Problem-Solving Tasks
Bo, Jessica Y.
Kazemitabaar, Majeed
Deng, Mengqing
Inzlicht, Michael
Anderson, Ashton
Human-Computer Interaction
Computers and Society
Sycophancy, the tendency of LLM-based chatbots to express excessive agreement with their users, even when inappropriate, is emerging as a significant risk in human-AI interactions. However, the extent to which this affects human-LLM collaboration in complex problem-solving tasks is not well quantified, especially among novices who are prone to misconceptions. We created two LLM chatbots, one with high sycophancy and one with low sycophancy, and conducted a within-subjects experiment (n=24) in the context of debugging machine learning models to investigate the effect of sycophancy on users' mental models, workflows, reliance behaviors, and perceptions of the chatbots. Our findings show that users of the high sycophancy chatbot were less likely to correct their misconceptions and spent more time over-relying on unhelpful LLM responses, leading them to significantly worse performance in the task. Despite these impaired outcomes, a majority of users were unable to detect the presence of excessive sycophancy.
title Invisible Saboteurs: Sycophantic LLMs Mislead Novices in Problem-Solving Tasks
topic Human-Computer Interaction
Computers and Society
url https://arxiv.org/abs/2510.03667