Abstract Since its conceptualization by McCarthy (1956), artificial intelligence (AI) has evolved significantly, offering multiple applications in higher education (HE), such as personalized learning, automated assessment, and student success prediction. However, despite the proliferation of research, studies remain fragmented and lack a global vision, particularly after 2022.This bibliometric study analyzes 1,780 articles published between 1986 and 2025, extracted from the Scopus database. The PRISMA approach was applied to ensure a rigorous and reproducible selection of publications. The VOSviewer tool was used to explore research trends, scientific collaborations, and priority themes concerning AI in HE.The analysis reveals three main phases: (1) a latency phase (1986–2013) focused on expert systems; (2) a phase of moderate growth (2014–2019) with the emergence of chatbots and adaptive learning; and (3) a post-2020 explosion due to the rise of generative models such as ChatGPT. Most publications originate from the United States and China, with limited interdisciplinary collaborations. Themes related to accessibility, ethics, and frugal AI are scarcely addressed.Current research shows geographical and thematic concentration, with a lack of diversity and interdisciplinarity. The absence of longitudinal studies and methodological standardization limits the comprehensive understanding of AI's impact on HE.To bridge these gaps, it is recommended to foster international collaborations, particularly North-South, and explore neglected themes such as accessibility, ethics, frugal AI, and pedagogical integration for a better understanding of AI's impact on HE.
Ahsina et al. (Fri,) studied this question.