Algorithmic academic framing in AI-driven smart libraries: a moderated mediation model of personalized learning and cognitive overload in academic decision-making.

This study examines how AI-driven smart libraries are associated with academic decision-making by introducing the concept of algorithmic academic framing. Drawing on cognitive load theory and social cognitive theory, a conditional process model was developed in which algorithmic academic framing is associated with academic decision-making quality through personalized learning, while AI self-efficacy and cognitive overload serve as boundary conditions. Time-lagged data were collected from 328 students across three universities in China. Structural equation modeling results indicate that algorithmic academic framing is positively associated with personalized learning, which, in turn, is positively associated with academic decision-making quality. The indirect association is strengthened when AI self-efficacy is high and weakened when cognitive overload is elevated. These findings suggest that AI-driven academic systems may function as cognitive infrastructures whose impact appears to depend on both user capability and cognitive resource availability. The study advances theoretical understanding of algorithmic influence in higher education and provides practical guidance for the design and implementation of AI-driven library platforms.