Artificial intelligence (AI) is rapidly transforming higher education, yet the structure and evolution of research in this field remain fragmented. This study provides a bibliometric analysis of scientific production on AI in higher education, aiming to identify key research trends, thematic areas, and emerging gaps. A total of 310 documents from Scopus and Web of Science were analyzed following PRISMA guidelines. Beyond mapping publication patterns, the study identifies dominant research lines, including academic integrity, teaching innovation, and digital transformation, as well as underexplored areas related to pedagogical integration and teacher training. The findings highlight a strong concentration of research in a limited number of thematic areas and geographical contexts, suggesting the need for more diverse and pedagogically grounded approaches. The study contributes to the field by outlining future research directions and supporting a more coherent development of AI in higher education.