When Recommendation Becomes Judgment: Identity Threat and Consumer Resistance to Generative AI Recommendations

Generative artificial intelligence (AI) recommendations are increasingly embedded in everyday consumption decisions, yet consumers are far from passive recipients. Drawing on fourteen open-ended questionnaires and a three-level coding procedure, this study explores how consumers perceive identity threat when facing generative AI recommendations and how they protect their sense of self-worth. The findings show that identity plays three roles in decision making: it acts as a pre-screening filter, as a reference point for evaluating consequences, and as a carrier of economic interests. Identity threat arises from three sources: the lack of situational knowledge in AI, the averaging logic of algorithms that erodes individual uniqueness, and a feeling of being watched that triggers anxiety about autonomy. In response, consumers adopt four strategies that form a continuum from passive withdrawal to active control: total avoidance, reverse selection, partial reference, and active steering. Based on these findings, the study proposes four theoretical propositions, namely identity precedence, context dependence, the strategy spectrum, and the resistance paradox, and discusses their implications for the design of recommender systems.