The Impact of AI-Enhanced Digital Textbooks on University Students’ Academic Performance in China

This research explores the mediating mechanisms and contingent factors that underlie the effectiveness of AI-enhanced digital textbooks (AI-EDT) for Chinese university students. Using partial least squares structural equation modeling (PLS-SEM) on data from 554 students, this paper proposes a conceptual framework where AI textbook features (AITF) affect academic performance (AP) via two distinct pathways: perceived learning impact (PLI) and self-directed learning (SDL). The results indicate that AITF exerts a positive effect on PLI (β = 0.427) and SDL (β = 0.356), which partially mediate the AITF-AP relationship. Learning infrastructure and support (LIS) moderates the AITF-PLI link (β = 0.189), while comparative effectiveness (CE) reinforces the PLI-AP connection (β= 0.234). This shows that AI-EDT’s success is not merely technological but contingent upon robust institutional ecosystems and learners’ favorable comparative perceptions. It provides an evidence-based framework for integrating AI-driven tools and highlights the synergy of pedagogical innovation, behavioral adaptation, and contextual support.