A field in fast-forward: thematic evolution of generative AI research in higher education through bibliometric analysis

Abstract

Generative artificial intelligence (GenAI) has rapidly become a transformative area of inquiry due to its potential to reshape higher education. However, existing bibliometric and review-based studies have often focused on specific tools or single databases, providing limited insight into the field’s broader intellectual structure. This study presents a bibliometric analysis of 3,888 publications retrieved from Web of Science and Scopus. The findings reveal an exceptional annual growth rate of 273.81%, with life-cycle modeling indicating that the field entered a late-growth phase in 2025 and is projected to reach early maturity by 2026. Publication output was concentrated in educational technology journals, while open-access venues demonstrated substantial publication activity and citation visibility. Geographically, research production was led by the United States and China, whereas many countries in Africa, the Caribbean, and Central Asia remained minimally represented or absent. International collaboration was limited and characterized by regional clustering across Anglophone, Iberian–Latin American, European, and Asian communities. Keyword co-occurrence analysis identified ChatGPT as the dominant organizing concept, with major thematic concentrations in academic integrity and assessment, governance and pedagogical challenges, AI literacy and ethics, critical thinking and feedback, technology adoption, and personalized learning. However, the strong focus on ChatGPT may limit the generalizability of findings across the broader GenAI ecosystem. Thematic mapping further showed that technology application and governance themes are comparatively well developed, whereas learner-centered concerns, including student experience, psychosocial factors, critical thinking, and instructor perspectives, remain less integrated.