AI가 어떻게 보험을 추천해야 고객이 받아들일까?

Addressing the disparity between the rapid proliferation of AI technology and the lagging consumer trust, this study empirically investigates how the interaction mode of AI-based insurance recommendation services influences users' psychological experiences and acceptance intentions. Specifically, the study analyzes the structural relationships in which the level of user involvement required by AI (Low involvement: Information-providing vs. High involvement: Co-design) affects acceptance intention mediated by the psychological needs of autonomy, competence, and relatedness. Furthermore, it examines the moderating effect of individual digital self-efficacy on these pathways. A scenario-based survey conducted with 400 participants nationwide revealed that the 'High Involvement' mode, which entails direct user interaction and contribution to outcome derivation, significantly enhances autonomy, competence, and relatedness compared to the simple information-providing mode. Notably, competence and relatedness were identified as key mechanisms mediating the relationship between the level of involvement and acceptance intention. Additionally, digital self-efficacy demonstrated a moderating effect by reinforcing the perceived autonomy and competence in high-involvement scenarios. The findings further indicated that 'relatedness' (emotional trust) serves as the primary driver of acceptance intention for the low-involvement group, whereas 'competence' (task understanding) is the critical determinant for the high-involvement group. This study empirically demonstrates that the level of user participation in AI service design induces complex psychological mechanisms beyond mere convenience. Consequently, it suggests the necessity of adaptive and personalized interface designs tailored to individual digital proficiency, rather than applying a uniform interface to all users.