Does greater chatbot usage lead to more positive affective attitudes toward AI? A moderated mediation model of relational capacity, competence, and privacy concerns in a two-wave panel study

Abstract

This study investigates the relationship between AI chatbot usage and affective attitudes toward AI, utilizing two-wave panel data with 579 and 391 participants, respectively. Integrating the CASA model, social cognitive theory, and privacy calculus theory, this research explores a moderated mediation model shaping affective attitudes toward AI. The results indicated that frequent interactions with AI chatbots were positively associated with positive affective attitudes, mediated by perceived relational capacity. Privacy concerns were negatively associated with affective attitudes toward AI. However, strong relational capacity could mitigate this effect, allowing users with high privacy concerns to maintain a positive affective attitude toward AI. This research enhances our understanding of privacy concerns in the AI era and underscores the importance of enhancing the relational capacity of chatbots while balancing the personalization-privacy dilemma when designing AI chatbots.