Trust Me on This: A User Study of Trustworthiness for RAG Responses

2026-01-20

AI Summary

This study investigates how structured explanations—specifically source attribution, factual grounding, and information coverage—affect user trust in synthetic answers generated by retrieval-augmented generation (RAG) systems. Through a two-stage user experiment, the research demonstrates that trust is not solely determined by objective answer quality but is significantly moderated by the clarity and actionability of explanations, as well as users’ prior knowledge. These findings challenge the prevailing accuracy-centric paradigm of credibility assessment in generative AI. The results further indicate that well-designed explanatory mechanisms can effectively guide users toward higher-quality responses; however, subjective factors continue to exert a strong influence on trust judgments, underscoring the complex interplay between system transparency and human perception in information retrieval contexts.

Technology Category

• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust
• Humans and AI: Explainable AI (XAI) for Human Understanding
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

The integration of generative AI into information access systems often presents users with synthesized answers that lack transparency. This study investigates how different types of explanations can influence user trust in responses from retrieval-augmented generation systems. We conducted a controlled, two-stage user study where participants chose the more trustworthy response from a pair-one objectively higher quality than the other-both with and without one of three explanation types: (1) source attribution, (2) factual grounding, and (3) information coverage. Our results show that while explanations significantly guide users toward selecting higher quality responses, trust is not dictated by objective quality alone: Users'judgments are also heavily influenced by response clarity, actionability, and their own prior knowledge.

Problem

• trustworthiness
• retrieval-augmented generation
• user study
• explanations
• information access

Innovation

• retrieval-augmented generation
• user trust
• explanations
• information access systems
• human-AI interaction