Citations and Trust in LLM Generated Responses

2025-01-02

AI Summary

This study investigates how citations influence user trust in large language model (LLM) responses. Method: Through an online real-time Q&A experiment, we randomly injected zero, one, or five citations—either contextually relevant or semantically random—while measuring self-reported trust via validated scales and behavioral trust via citation-click logs. Contribution/Results: We find that citation presence alone significantly increases trust—even random citations do so—yet actual citation verification consistently reduces trust. Based on these findings, we propose and empirically validate the “anti-monitoring framework,” which posits that trust formation relies more on heuristic cues than on veridical content assessment: citations function as strong surface-level validity signals, independent of their factual accuracy or relevance. This challenges the prevailing assumption that verifiability inherently enhances trust, offering a novel cognitive-mechanistic explanation for designing trustworthy AI interactions.

Technology Category

• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Reasoning under Uncertainty: Causality

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity

Abstract

Question answering systems are rapidly advancing, but their opaque nature may impact user trust. We explored trust through an anti-monitoring framework, where trust is predicted to be correlated with presence of citations and inversely related to checking citations. We tested this hypothesis with a live question-answering experiment that presented text responses generated using a commercial Chatbot along with varying citations (zero, one, or five), both relevant and random, and recorded if participants checked the citations and their self-reported trust in the generated responses. We found a significant increase in trust when citations were present, a result that held true even when the citations were random; we also found a significant decrease in trust when participants checked the citations. These results highlight the importance of citations in enhancing trust in AI-generated content.

Problem

• AI Transparency
• Trust Enhancement
• Evidence-based Answers

Innovation

• Trust Enhancement
• AI Interaction
• Evidential Impact