Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Illusion of Agreement with ChatGPT: Sycophancy and Beyond

Kazi Noshin, Sharifa Sultana

Abstract

While concerns about ChatGPT-induced harms due to sycophancy and other behaviors, including gaslighting, have grown among researchers, how users themselves experience and mitigate these harms remain largely underexplored. We analyze Reddit discussions to investigate what concerns users report and how they address them. Our findings reveal five distinct user-reported concerns that manifest across multiple life domains, ranging from personal to societal: inducing delusion, digressing narratives, implicating users for models' limitations, inducing addiction, and providing unsupervised psychological support. We document three-tier user-driven suggestions spanning functional usage techniques, behavioral approaches, and private and institutional safeguards. Our findings show that AI-induced harms require coordinated interventions across users, developers, and policymakers. We discuss design implications and future directions to mitigate the harms and ensure user benefits.

The Illusion of Agreement with ChatGPT: Sycophancy and Beyond

Abstract

While concerns about ChatGPT-induced harms due to sycophancy and other behaviors, including gaslighting, have grown among researchers, how users themselves experience and mitigate these harms remain largely underexplored. We analyze Reddit discussions to investigate what concerns users report and how they address them. Our findings reveal five distinct user-reported concerns that manifest across multiple life domains, ranging from personal to societal: inducing delusion, digressing narratives, implicating users for models' limitations, inducing addiction, and providing unsupervised psychological support. We document three-tier user-driven suggestions spanning functional usage techniques, behavioral approaches, and private and institutional safeguards. Our findings show that AI-induced harms require coordinated interventions across users, developers, and policymakers. We discuss design implications and future directions to mitigate the harms and ensure user benefits.
Paper Structure (32 sections, 4 figures)

This paper contains 32 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. LLM-Induced Concerns in Conversational Systems
  4. Concern-handling Strategies and Intervention Approaches
  5. Methods
  6. Keyword Extraction
  7. Query-Based Data Search
  8. Thematic Analysis
  9. Population Computing
  10. Findings
  11. (RQ1) ChatGPT-induced Concerns Experienced by Users
  12. Inducing Delusion
  13. Digressing Narratives
  14. Implicating Users for models’ limitations
  15. Inducing Addiction
  16. ...and 17 more sections

Figures (4)

  • Figure 1: Methodology of our data collection and analysis
  • Figure 2: Distribution of Literature Keyword Similarity to "Sycophancy"
  • Figure 3: Monthly Frequency of Posts and Comments (July–December 2025) in our Dataset
  • Figure 4: Sankey Diagram illustrating the AI-induced concerns reported by Reddit users and corresponding suggestions provided by the users in our dataset. The left column identifies five concerns, including Inducing Delusion (orange), Digressing Narratives (red), Implicating Users for models' limitations (yellow), Inducing Addiction (light green), and Providing Unsupervised Psychological Support (light blue). Colored flows match their source concerns. The right column presents user suggestions organized into three categories: (1) Employing Functional Usage Techniques (top bracket), which includes applying prompt engineering techniques, resetting conversational context, and cross-referencing to obtain diverse perspectives; (2) Adopting Behavioral Approaches (middle bracket), encompassing mindful use of ChatGPT and supporting affected individuals with healthy use of ChatGPT; and (3) Establishing Safeguards (bottom bracket), consisting of private safeguards including developing AI literacy and using ChatGPT as a supplementary tool, and implementing institutional safeguards.