Influence of Agentic Capability of AI Chatbots on Value Co-Creation and Self-Directed Learning Ability -A Dual Moderated Mediation Model via Critical AI Literacy and AI Self-Efficacy-

초록

Library AI chatbots have evolved from information-retrieval tools into agentic systems capable of autonomous reasoning and adaptive interaction, substantially expanding their educational potential. This study examines how the agentic capability of university library AI chatbots affects students' self-directed learning ability through value co-creation as a mediating mechanism, with critical AI literacy and AI self-efficacy as stage-specific moderators. An online survey was administered to 434 students at four top-tier Chinese university with direct experience using library AI chatbot services. The statistical analysis for the dual moderated mediation model demonstrated that agentic capability exerts a significant positive direct effect on self-directed learning ability, with value co-creation functioning as a partial mediator. Critical AI literacy moderated the agentic capability value co-creation pathway as a threshold gateway condition, while AI self-efficacy moderated the value co-creation to self-directed learning ability pathway as a continuous amplifier. The conditional indirect effect was non-significant when both moderators were low but became significant and substantial when both were high, indicating that the indirect pathway is activated only once both moderators reach sufficient levels. These findings extend Service-Dominant Logic to AI-mediated educational contexts and reveal a fundamentally asymmetric, sequentially gated moderation structure that repositions self-directed learning ability as a relational construct co-shaped by agentic AI properties and learners' cognitive and psychological resources.

키워드