University students’ overreliance on generative AI feedback in the learning process: the effects of perceived cognitive load, cognitive offloading, and metacognitive monitoring

Generative artificial intelligence (GAI) has introduced an important question in education, when AI provides immediate and reliable feedback, do students use it as a learning support tool or gradually rely on it as a substitute for their own cognitive efforts? Drawing on Cognitive Load Theory, Cognitive Offloading Theory, and Metacognitive Monitoring Theory, this study develops a theoretical model to explain the formation process of undergraduate students’ overreliance on GAI feedback. Specifically, this study examines the relationships among perceived cognitive load, cognitive offloading, and overreliance on GAI feedback, while further exploring how metacognitive monitoring levels shape this process. A questionnaire survey was conducted among 600 undergraduate students, and structural equation modeling was performed using AMOS 31.0. The results showed that perceived cognitive load significantly and positively predicted cognitive offloading, which further contributed to students’ overreliance on GAI feedback. Meanwhile, perceived cognitive load also directly predicted overreliance on GAI feedback. Bootstrap analysis further confirmed that cognitive offloading mediated the relationship between perceived cognitive load and overreliance on GAI feedback. In addition, this study established separate structural equation models for students with different levels of metacognitive monitoring and found distinct patterns in the formation of GAI feedback overreliance. Specifically, students with lower metacognitive monitoring were more likely to be influenced by cognitive pressure during learning, showing a more prominent direct relationship between perceived cognitive load and GAI feedback overreliance. This finding suggests that these students tend to rely more directly on GAI feedback when facing higher cognitive demands. In contrast, students with higher metacognitive monitoring demonstrated a more evident cognitive regulation process, in which cognitive offloading played a more important role in linking perceived cognitive load with GAI feedback overreliance. This indicates that these students may actively adjust their cognitive resource allocation when using GAI, although continuous cognitive offloading may still increase their dependence on AI-generated feedback.