Psychological Adaptation and Affective Feedback in AI-Enhanced Education: A Systematic Review

Abstract

This review synthesizes current research at the intersection of artificial intelligence (AI), education, and psychology, with a particular focus on the psychological mechanisms underlying learner adaptation. Evidence from affective computing illustrates the pedagogical value of multimodal emotion recognition, while meta-analyses of large language model (LLM)-based tutoring systems reveal both the potential for enhanced engagement and the risks of cognitive dependency. Empirical studies on adoption further highlight the mediating roles of self-efficacy, motivation, and anxiety in sustaining learner interaction with AI tools. Case analyses identify progressive adaptation, emotional regulation, and transparency as critical enablers of effective learning experiences. The review concludes that ethically grounded, human-centered AI design is indispensable for cultivating resilient, adaptive, and learner-centered educational ecosystems.