Artificial intelligence in science education with a focus on life and earth sciences a bibliometric analysis of global research trends from 2015 to 2025

Abstract

Given the sustained and extraordinary growth of scientific output, it is crucial to map in detail its intellectual structure and its evolutionary processes. This research provides a detailed bibliometric analysis of AI in education publications using Scopus indexed data from 2015 to 2025. We analyzed a corpus of 779 publications through a set of quantitative indicators and structural analyses, including the evolution over time of the publications, the main editorial sources, the profiles of the authors, geographical and institutional contributions, funding organizations and networks of co-occurrence of keywords. The visualisations were produced with VOSviewer and allowed to identify relational configurations and emerging thematic hubs. The results show an exponential growth curve from one publication in 2015 to 378 in 2025 with a significant inflection point beginning in 2023. Publication growth accelerated markedly after 2023, temporally coinciding with the widespread emergence of generative AI tools such as ChatGPT; while the bibliometric design cannot establish a causal link, this coincidence is consistent with a broader reorientation of research priorities toward generative AI in the discipline. This acceleration also raises questions about the sustainability of this growth and the risk of a trend effect that could influence the quality and depth of contributions. In terms of geography, the United States is leading with the number of publications (n = 168) and funding, mostly by the National Science Foundation, confirming their structuring position in the orientation of scientific agendas. However, the emergence of countries like China, the United Kingdom and Spain suggests a gradual internationalisation of the field, though asymmetrical in terms of access to resources and innovation capacity. Computers and Education: Artificial Intelligence and Sustainability (Switzerland) excel, respectively, in influence and productivity at the journal level. But this editorial concentration could be the sign of a structuring of the field that is beginning to emerge, with a reliance on a few dissemination channels. Finally, the papers by Dragan Gašević and Kenneth R. Koedinger show the centrality of some researchers in the consolidation of the field, while indicating a possible concentration of scientific influence. In sum, this study illustrates the fast-paced growth of the field and the structural tensions that arise in this process, paving the way for future studies that seek to deepen the qualitative, ethical, and pedagogical dimensions of AI integration in education.