FROM MAPPING TO THEORY: UNCOVERING THE INTELLECTUAL STRUCTURE AND CONCEPTUAL EVOLUTION OF GENERATIVE ARTIFICIAL INTELLIGENCE RESEARCH IN HIGHER EDUCATION

Abstract

The rapid emergence of Generative Artificial Intelligence (GenAI) has transformed higher education by reshaping teaching, learning, assessment, and institutional practices. Despite the rapid growth of scholarly publications, existing bibliometric studies have largely focused on descriptive indicators, providing a limited understanding of the field's intellectual foundations, conceptual evolution, and theoretical development. This study addresses this gap by conducting a comprehensive bibliometric and science mapping analysis of 3,214 Scopus-indexed publications on Generative AI in higher education published between 2020 and 2026. Following the PRISMA 2020 framework, bibliographic data were analyzed using VOSviewer version 1.6.20 to examine publication trends, co-citation networks, and keyword co-occurrence patterns. The findings reveal exponential growth in research following the widespread adoption of large language models, with publication output expanding rapidly since 2023. Co-citation analysis identified five major intellectual domains encompassing AI-assisted learning applications, student engagement, methodological and theoretical foundations, AI-supported language learning, and technology acceptance. Co-word analysis further revealed four dominant conceptual themes: pedagogical integration, learner-centered research, computational technologies, and ethical governance. Collectively, these findings demonstrate that Generative AI research has evolved from technologyoriented investigations to a multidisciplinary educational ecosystem that emphasizes pedagogical innovation, institutional transformation, and responsible AI implementation. Building upon these empirical findings, the study proposes the AI Pedagogical Knowledge Ecosystem Framework, which integrates technological innovation, pedagogical transformation, learner engagement, institutional governance, and educational outcomes into a unified conceptual model. By combining co-citation and co-word analyses within a single science mapping framework, this study extends previous bibliometric research beyond descriptive mapping toward theory-informed conceptual development and knowledge synthesis. The findings provide valuable guidance for educators, higher education institutions, policymakers, instructional designers, and researchers seeking to support the responsible and sustainable integration of Generative AI in higher education.