A bibliometric analysis of agentic AI and the agentic pedagogical agency framework

Abstract

The rapid evolution of generative artificial intelligence (GenAI)  has given rise to agentic artificial intelligence (AI) systems capable of autonomous planning, reasoning, and goal-oriented action, positioning education at the center of a profound pedagogical transformation. Despite growing interest, research on agentic      (AI) in education remains conceptually fragmented and unevenly theorized. This study aims to systematically map the intellectual, conceptual, and collaborative structures of the Agentic AI and Education literature and to advance a pedagogically grounded framework for understanding agency in AI-mediated learning environments. Using a comparative bibliometric research design, publications indexed in the Web of Science Core Collection (WoS)  and Scopus were analyzed through keyword co-occurrence, co-citation, bibliographic coupling, and co-authorship networks visualized with VOSviewer. The findings reveal a clear epistemic divergence between databases: while WoS      emphasizes technological legitimacy, ethics, and governance, Scopus foregrounds applied pedagogical contexts, learner autonomy, and human–AI collaboration. Across both datasets, agentic AI is predominantly framed through technological constructs, with pedagogical agency remaining underarticulated. To address this gap, the study proposes the Agentic Pedagogical Agency Framework (APAF), conceptualizing agency as a relational and dynamic construct distributed among learners, teachers, agentic AI systems, and educational institutions across shared, delegated, and negotiated modes. By integrating bibliometric evidence with theoretical synthesis, this study contributes a novel analytical lens for interpreting agentic AI in education and offers directions for research, practice, and governance in emerging AI-mediated learning ecosystems. Rather than reviewing all educational applications of agentic AI, this study centers on agency as the core analytical construct shaping AI-mediated education.