Uncovering the Mediating Effects of <scp>AI</scp> Trust and Self‐Efficacy on Engagement in Human‐ <scp>GenAI</scp> Communication for <scp>EFL</scp> Learning
Qinqing Zhang, Xiaoquan Pan, Jiqun Fan
Abstract
ABSTRACT Generative Artificial Intelligence (GenAI) offers robust information generation, swift responsiveness and high personalization for education. However, it remains unclear how these specific technological features translate into improved learning engagement through the mediation of learners' psychological states. Grounded in the Stimulus‐Organism‐Response (S‐O‐R) theory, this study investigates the mediating roles of AI trust and AI learning self‐efficacy in the relationships between GenAI features (responsiveness and personalization) and the learning engagement of English as a Foreign Language (EFL) learners. This study surveyed 697 undergraduate students from four universities in China, all of whom had used GenAI to assist their EFL learning. Using structural equation modelling, the results revealed that: (1) Both responsiveness and personalization significantly enhanced learners' AI trust and self‐efficacy; (2) These psychological states, in turn, positively influenced learning engagement; (3) A critical divergence was observed: responsiveness influenced engagement only indirectly through the full mediation of trust and self‐efficacy, whereas personalization exerted both a direct effect and indirect effects through the same mediators. This study elucidates the distinct psychological pathways through which different AI features affect engagement, providing a more nuanced theoretical account. The findings underscore that for AI tools to foster deep engagement, they must be designed not only to be responsive and personalised but also to proactively build learners' trust and confidence in their AI‐use capabilities.