Meta-synthesis and bibliometric analysis of the use of artificial intelligence in science education

ABSTRACT This study aims to systematically examine the scientific implications of artificial intelligence (AI) technologies in the context of science education and to analyze the academic production generated at the intersection of these two fields. As 21st century skills and digital transformation become increasingly important in educational systems, AI-supported applications offer significant opportunities for disciplines that prioritize cognitive skills, such as science education. The potential of AI is particularly noteworthy in terms of creating personalized learning environments, monitoring student performance in real time, and dynamically adapting learning processes. The research utilized two different methods based on both international and national academic publications: bibliometric analysis and meta-synthesis. The bibliometric analysis was conducted on 471 publications obtained through comprehensive searches of the Web of Science (WoS) and Scopus databases using the keywords "science education" and "artificial intelligence." The analysis included detailed analyses of publication distribution by year, the most cited studies, the most influential authors, institutions, and countries, as well as keyword trends and scientific collaborations. In this context, the field's development trajectory, focused research themes, and research gaps were identified. The second method of the research, meta-synthesis, conducted an in-depth analysis of recently published theses and academic articles in Turkey. In this context, 22 studies (15 theses, 7 articles) located in the National Thesis Center and DergiPark databases were systematically examined. The meta-synthesis method comparatively evaluated the research objectives, methods used, data collection tools, sample groups, and results of these studies. This provided a detailed overview of the level of integration between science education and AI in the national context and the areas of focus. The analyses revealed that AI applications in science education generally revolve around the following themes: intelligent tutoring systems, automated assessment systems, augmented and virtual reality-based applications, learning analytics, the educational use of natural language processing systems such as ChatGPT, and metaverse-supported science education environments. These applications were found to have particularly positive effects on conceptual understanding, scientific process skills, personalized learning, and student motivation. Additionally, it has been determined that the number of publications on AI in science education in Turkey has increased significantly in recent years, but methodological diversity is limited and mostly consists of descriptive studies. Internationally, countries such as the United States, China, and the United Kingdom have been found to be pioneers in this field, with interdisciplinary studies particularly focused on STEM, educational technologies, and computer science. Concept maps and collaboration networks created through open-source software such as VOSviewer and Bibliometrix have provided the opportunity to visualize the structural aspects and contextual focal points of scientific production. In conclusion, this thesis provides significant contributions to the body of knowledge in the field by examining the relationship between science education and AI technologies at both macro and micro levels. The findings are a comprehensive resource for both researchers and institutions developing educational policies. Furthermore, methodological limitations, practical shortcomings, and research gaps encountered in the integration of AI into science education have been identified, and recommendations for future research have been developed. In this context, the research not only describes the current situation; It also aims to guide future academic and pedagogical studies. Keywords: Science Education, Artificial Intelligence, Bibliometric Analysis, Meta-Synthesis, Educational Technologies.