“When AI lies to me”: Exploring consumer perceptions of AI hallucination in marketing contexts

Absztrakt (kivonat)

Generative artificial intelligence (GAI) has rapidly emerged as a consequential mediating technology in consumer decision-making, yet its propensity to produce fluent, confident, and factually erroneous outputs, commonly termed AI hallucinations, poses significant implications for consumer welfare and brand trust. This study investigates how consumers perceive, interpret, and behaviorally respond to AI hallucinations encountered during marketing-relevant tasks, including product research, price comparison, and travel planning, across widely adopted platforms such as ChatGPT, Gemini, and brand-integrated chatbots. To examine this question, an exploratory qualitative study was conducted, drawing on 30 semi-structured in-depth interviews with consumers who had personally encountered AI hallucinations in purchase- related contexts, with data analyzed inductively using NVivo. The analysis yields eight interrelated concepts organized across four dimensions. Cognitively, fluency-credibility bias captures how hallucinated outputs gain credibility through authoritative linguistic form, while heuristic over-reliance describes how speed and convenience suppress critical evaluation. Affectively, expectancy disconfirmation produces frustration and perceived betrayal upon hallucination detection, while affective habituation characterizes the emotional dampening among more experienced users. Relationally, intentionality attribution reflects consumers’ negotiation of whether AI failures constitute deception, and asymmetric accountability documents the systematically different standards applied to AI versus human errors. Behaviorally, compensatory prompting describes deliberate interaction modifications developed in response to prior hallucinations, while calculative continuance captures continued use despite recognized limitations. Together, these dimensions are interpreted through the Trust Recalibration Cycle, an integrative theoretical contribution describing how consumers progressively redefine the role AI is permitted to play in their decision-making environment. These findings contribute to emerging literature at the intersection of AI-mediated communication, consumer trust, and technology adoption, with substantive implications for the responsible design and governance of AI-powered marketing tools.