Understanding perceptual-AI dissonance: A conceptual framework for human–AI interaction in hospitality

This study elaborates on the notion of AI behaviors and conceptualizes perceptual-AI dissonance. The study proposes that customers respond to AI service agents based on their perceptions of AI behavior rather than the underlying AI systems. A cognitive gap between actual AI behaviors and perceived AI behaviors leads to perceptual-AI dissonance, which is a key driver of human-AI interaction outcomes. The study presents a typology of AI’s behavior, identifying four types: adaptation, imitation, normative alignment, and rational decision-making. Based on this typology, a series of in-depth interviews was conducted, revealing nine types of perceptual-AI dissonance in hospitality service contexts. Additionally, three antecedents of perceptual-AI dissonance were identified. The Perceptual-AI Dissonance (PAID) framework was proposed. By integrating computer science and social science approaches, this study offers a novel and often overlooked perspective for understanding human-AI interaction through the conceptual framework of perceptual-AI dissonance. Theoretical and practical implications are discussed.