AI-Driven Personalized Learning and the Transformation of Secondary Education: Pedagogical Models, Cognitive Impact, Equity Implications, and Implementation Frameworks (2020–2026)

Recent advances in artificial intelligence (AI) have enabled large-scale deployment of adaptive learning platforms, intelligent tutoring systems, and learning analytics in secondary education, promising unprecedented levels of personalization while raising new pedagogical, cognitive, and ethical questions. Meta-analytic evidence from 2020–2026 indicates that AI-assisted personalized learning yields overall moderate positive effects on student achievement and affective outcomes, with effect sizes comparable to other well-established instructional interventions. At the same time, studies document heterogeneous impacts moderated by implementation context, teacher orchestration, and students’ prior attainment, as well as persistent concerns regarding algorithmic bias, data privacy, and the digital divide. This conceptual paper synthesizes recent empirical and theoretical work to examine how AI-driven personalization intersects with major pedagogical models (constructivism, competency- based learning, inquiry- and project-based learning) and core learning theories (Vygotsky’s zone of proximal development, self-determination theory, cognitive load theory, mastery learning) in secondary schooling. It reviews evidence on cognitive and psychological effects for adolescents, including attention, motivation, reward sensitivity, and self-efficacy, drawing on meta-analyses of AI- enabled learning and teacher–AI collaboration. Equity and ethics are analyzed through emerging frameworks of algorithmic fairness, cyclical digital divides, and regulatory regimes such as FERPA and GDPR. Building on cross-national developments in Finland, Singapore, the United States, and Canada, the paper proposes the AI-Augmented Secondary Education Implementation Model (AASEIM), a multi-layer framework encompassing governance, technology, pedagogy, psychological safeguards, and evaluation. Conceptually, AI-driven personalization is argued to be most beneficial when embedded in human-in-the-loop, pedagogy-first designs that support teacher agency, safeguard youth cognition and wellbeing, and explicitly address structural inequities.