Analysis Group Authors Examine Consumers’ Skepticism Toward, Verification of, and Reliance on Hallucinated Information from AI Advisors

October 5, 2026

Organizations increasingly rely on chat bots and other conversational AI agents to assist users with tasks such as obtaining health advice or searching for a consumer product. As generative AI systems have the capacity to provide inaccurate responses – often referred to as hallucinations – it is important to understand the factors that may impact whether consumers check or rely on hallucinated responses from conversational AI agents.

In an article published in Current Opinion in Psychology, Principal Laura O’Laughlin, academic affiliate Simon Blanchard, and Professor Aaron Garvey examine the current literature on goal-directed interactions between individuals and conversational AI agents that involve risk of hallucinations. They present findings related to three separate constructs that are frequently conflated in the existing body of research: user skepticism (the recognition that a response is incorrect or may require verification), verification (the use of one or more sources to check the accuracy of the response), and reliance (the use of the response in the user’s decision). The authors discuss organizational and policy responses to hallucination risk, including mitigation efforts to reduce frequency of hallucinations and interventions to shape users’ responses. The article identifies several avenues for future research, including understanding individual differences in skepticism and reliance and exploring ways to improve user recognition of higher-risk content categories.

Read the article
Learn more about how AG’s expertise is amplified by AI