Research trends in generative AI in higher education: A bibliometric and systematic review approach
Abstract
Generative AI (GenAI) has transformed teaching and learning in higher education, with publication volumes surging annually. However, previous bibliometric reviews focus either on general AI or ChatGPT, lacking a specific mapping of GenAI publication trends and the characteristics of highly impactful articles in higher education, a gap that this study addresses. This study aims to: (1) map the change in the number of GenAI research publications in higher education over time; (2) identify emerging research trends through co-occurrence network analysis; and (3) analyse the characteristics of highly cited GenAI articles in higher education. A systematic review following PRISMA guidelines was conducted, identifying 913 relevant articles from Scopus and Web of Science (2023–2025). Bibliometric analysis, co-occurrence network analysis, and an in-depth examination of the top 2% most-cited articles (n=18) were employed. Regarding the first aim, the findings reveal an annual publication growth rate of 332.4%, with 33 articles in 2023, 263 in 2024, and 617 in 2025. Regarding the second aim, co-occurrence network analysis identifies three emerging trends: (1) pedagogical shifts and ethical challenges (academic integrity, AI literacy, critical thinking), (2) psychological and demographic factors influencing student attitudes towards GenAI, and (3) technology adoption models (TAM, UTAUT). Regarding the third aim, among highly cited articles, 50% used mixed methods, 27.78% involved international collaboration, and 72.22% focused on pedagogical transformation and AI literacy, while student demographics (11.11%) and technology acceptance frameworks (22.22%) remain underexplored. This research indicates a shift from technological exploration towards a more interdisciplinary approach integrating ethical, psychological, and pedagogical perspectives. These findings provide actionable insights for researchers, educators, and policymakers to prioritise collaborative, mixed-methods research and address gaps in demographic and adoption-model studies to advance ethical and effective GenAI integration in higher education.