System, User, and Design Factors in AI Tutor Engagement

This study investigates how teacher-like and peer-like AI tutors influence student learning engagement (LE) through system and user characteristics. Building on the information systems success model and learning engagement theory, we propose the AI Tutor Engagement Framework (AITEF) to explain how content quality, platform usability, feedback quality, learning self-efficacy, and learning innovativeness contribute to LE. An experimental design compared two AI tutor prototypes: one adopting a teacher-like persona and the other a peer-like persona. Data were analyzed using PLS-SEM and fsQCA. Results demonstrate that all system and user factors significantly enhance LE, and tutor persona moderates several structural relationships. fsQCA further identifies multiple causal pathways leading to high engagement, emphasizing equifinality. The findings extend research on human–AI interaction in education and offer practical guidance for designing AI tutors that foster meaningful and sustained student engagement.