Human-Centered Education in the Age of Artificial Intelligence: A Delegation Framework for Teacher–Student Relationships, Learner Agency, and Ethical Formation

Abstract

As artificial intelligence (AI) systems become increasingly capable of performing instructional, explanatory, assessment, feedback, and tutoring functions, education faces a question that is theoretical as much as technological: which educational functions can appropriately be delegated to AI, and which require continuing human participation because their value depends on relational, agentic, ethical, or accountable forms of educational practice? This article develops the Human-Centered AI Education Framework (HCAI-EF), a theoretical framework for addressing this delegation problem. Drawing on relational pedagogy, theories of subjectification and educational purpose, social-cognitive and self-determination perspectives on agency, and dialogic and care traditions, HCAI-EF conceptualizes human-centered AI-mediated education through three interdependent dimensions: teacher–student relationships, learner agency, and ethical formation. The framework distinguishes three modes of educational delegation: AI-delegable functions, for which automation does not ordinarily alter the central educational purpose of the activity; AI-augmented functions, in which AI may contribute substantially but meaningful human judgment and oversight remain necessary; and human-primary relational functions, for which interpersonal recognition, contextual judgment, moral accountability, or formation constitutes part of the educational value itself. Rather than classifying technologies categorically, the framework evaluates educational functions according to their relational constitutiveness, judgmental complexity, accountability requirements, and consequences for learner agency. It further distinguishes AI replacing educational tasks, augmenting teachers, augmenting learners, and mediating teacher–student interaction, recognizing that AI can expand accessibility, personalization, participation, and autonomy while also creating risks of cognitive substitution, inappropriate reliance, relational displacement, and diminished accountability. The article derives empirically testable propositions and specifies boundary conditions for future research. HCAI-EF therefore shifts the central question from whether AI can perform an educational task to whether, and under what conditions, delegating that task preserves the human purposes through which it acquires educational meaning.