The aim of this study is to conduct a comprehensive analysis of scientific studies on the sustainability of artificial intelligence (AI) in education in the 21st century, based on the Web of Science (WoS) database, using systematic and bibliometric methods. In this context, a total of 237 articles on the sustainability of AI in education, covering a historical perspective, have been included in the analysis. The retrieved documents have been analysed under five main categories: content analysis, scientific productivity, network analysis, conceptual structure and thematic mapping. Within this scope, scientific articles have been analysed through content analysis regarding research method types, research designs, theoretical and conceptual frameworks, preferred AI technologies, usage purposes and roles. Subsequently, scientific productivity, citation performance, collaboration networks, conceptual structure, thematic maps and intellectual transformations have been systematically and comprehensively evaluated. Throughout this process, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines have been followed as the basis for conducting the study. The findings indicate that scientific productivity in AI sustainability within education has significantly increased since 2019. China and the USA lead in academic production, whereas the United Kingdom, Germany and Spain play key roles in international collaborations. Most studies have been shaped within the framework of cognitive learning theories and educational technologies, with AI usage in personalised learning becoming increasingly prevalent. Research trends reveal a transition from digital learning to AI-driven educational systems and sustainable development goals.