Novelty or eeriness? Investigating consumer responses to AI-generated product images through expectancy violation theory

Purpose Drawing on Expectancy Violation Theory, this study examines how product image creator identity (AI vs. human) influences purchase intentions and the underlying psychological mechanisms. Design/methodology/approach Three online experiments examined consumers' attitudes toward AI-generated product images and purchase intentions toward the featured products, assessed how these evaluations deviated from expectations associated with human-created images, and identified conditions under which artificial intelligence (AI) imagery was more or less effective in shaping consumer responses. Data were analyzed in Statistical Package for the Social Sciences using Multivariate Analysis of Variance and Multivariate Analysis of Covariance, with serial and moderated mediation analyses conducted using the PROCESS macro. Findings Consumers consistently prefer human-created product images and perceive them as more novel, whereas AI-generated images evoke greater eeriness. Perceived novelty, eeriness, positive and negative expectancy violations, and attitudes mediate the relationship between image creator identity and purchase intention. Product type serves as a boundary condition, with AI-generated images evaluated more favorably for search products than for experience products. Originality/value This research identifies positive and negative expectancy violations as central psychological mechanisms shaping consumer responses to AI-generated product images. By highlighting product type as a boundary condition, it helps reconcile mixed findings in the AI-generated marketing literature and provides actionable guidance for the strategic use of AI-generated product imagery.