Keeping it fresh : what attributes of AI recommender systems do consumers value?

Abstract

Purpose – This study examines how consumers evaluate AI-powered retail recommender systems, focusing on key system attributes and their effects on consumer gratification and behavioral intention. In addition, the impact of consumer expertise as a moderator of these effects under different recommendation framing conditions is explored. Design/methodology/approach – In Study 1, we conducted an online survey (N = 435) and structural equation modeling to test how four AI recommender attributes–accuracy, diversity, novelty and serendipity–influence three types of consumer gratification and, in turn, behavioral intention. In Study 2, we employed a 2 (framing: accuracy vs. serendipity) × 2 (expertise: low vs. high) between-subjects experiment (N = 204) to examine how these variables affect consumer responses. Findings – All four AI attributes positively influence consumer gratification, which subsequently increases behavioral intention. Accuracy, diversity and serendipity exert broader effects across all gratification types, while novelty mainly enhances entertainment and interactivity. A significant interaction also emerged: whereas high-expertise consumers responded more positively to serendipity-based recommendations, low-expertise consumers preferred accuracy-based recommendations. Originality/value – By investigating the emerging attributes of AI systems, this study extends the application of affordance theory and the uses and gratifications (U&G) framework to AI-enabled retail settings. Identifying consumer expertise as a critical boundary condition, it reveals how and for whom AI recommendations generate the most value in interactive marketing.