Agentic AI Empowering Personalized Learning: Role Recon-struction and Pathway Optimization of Intelligent Agents in Educational Contexts

Abstract

The transition from generative artificial intelligence to agentic AI represents a critical paradigm shift in educational technology. Focusing on personalized learning in adult higher education, this paper integrates sociocultural learning theory, human-computer interaction principles, and big data analytics to investigate the role reconstruction and pathway optimization of educational agents. We clarify four core capabilities of agentic AI, propose a typology of educational agent roles, and illustrate three application scenarios: Socratic questioner, collaborative information hub, and affective support monitor. We further analyze key challenges including the metacognitive participation paradox, data privacy, algorithmic bias, and black-box opacity. Grounded in a human-AI augmentation paradigm, we provide targeted mitigation strategies and multi-stakeholder recommendations. This study offers theoretical and practical insights for the integration of AI, big data analytics, and social sciences in educational transformation, supporting the realization of UN SDG4.