Against the backdrop of the rapid diffusion of Artificial Intelligence Generated Content (AIGC) and the rise of the “surprise economy,” trendy blind-box toys have emerged as a consumption context characterized by emotional stimulation and community-based diffusion. Focusing on the question of “conformity or identification,” this study integrates the Stimulus-Organism-Response (SOR) framework and the Theory of Planned Behavior (TPB) to construct a comprehensive model. The model incorporates Psychological Expectation (PE)—including Novelty (NO), Scarcity (SC), Aesthetic Perception (AP), Emotional Value (EV), and Product Identification (PID)—as well as AI Creation Perception (AICP) and Conformity Psychology (CP), with Community Belongingness (CB) introduced as a moderating variable. Based on 514 valid responses, Structural Equation Modeling (SEM) and Artificial Neural Network (ANN) analyses were conducted. The results indicate that NO, AP, EV, PID, AICP, and SC significantly enhance Subjective Norm (SN). With the exception of SC, these variables, together with CP, also significantly strengthen Perceived Behavioral Control (PBC). Both SN and PBC positively influence Attitude (AT), with SN exerting a stronger effect. Furthermore, AT, SN, and PBC all significantly promote Purchase Intention (PI), among which AT demonstrates the greatest impact. CB significantly and positively moderates the relationships between SN and PI, as well as between PBC and PI. The ANN results further highlight the critical roles of NO, PID, and CP along key causal paths. These findings suggest that consumers’ PI are jointly driven by community norms and conformity tendencies, as well as by aesthetic preferences and psychological identification with AI-generated content. Theoretically, this study integrates PE, technology perception, and community mechanisms, supporting 20 of the 21 proposed hypotheses. Practically, it proposes conversion strategies centered on leveraging AIGC to enhance novelty-driven aesthetics, strengthen IP identification, and optimize community operations.
Yang et al. (Wed,) studied this question.