Abstract
The role of Artificial Intelligence (AI) in education is undergoing a rapidtransformation, moving beyond its historical function as an instructional tooltowards a new potential as an active participant in the learning process. Thisshift is driven by the emergence of agentic AI, autonomous systems capable ofproactive, goal-directed action. However, the field lacks a robust conceptualframework to understand, design, and evaluate this new paradigm of human-AIinteraction in learning. This paper addresses this gap by proposing a novelconceptual framework (the APCP framework) that charts the transition from AI asa tool to AI as a collaborative partner. We present a four-level model ofescalating AI agency within human-AI collaborative learning: (1) the AI as anAdaptive Instrument, (2) the AI as a Proactive Assistant, (3) the AI as aCo-Learner, and (4) the AI as a Peer Collaborator. Grounded in socioculturaltheories of learning and Computer-Supported Collaborative Learning (CSCL), thisframework provides a structured vocabulary for analysing the shifting roles andresponsibilities between human and AI agents. The paper further engages in acritical discussion of the philosophical underpinnings of collaboration,examining whether an AI, lacking genuine consciousness or sharedintentionality, can be considered a true collaborator. We conclude that whileAI may not achieve authentic phenomenological partnership, it can be designedas a highly effective functional collaborator. This distinction has significantimplications for pedagogy, instructional design, and the future research agendafor AI in education, urging a shift in focus towards creating learningenvironments that harness the complementary strengths of both human and AI.