A multidimensional review of artificial intelligence in science education examining trends, curriculum integration, and ethical implications

Abstract

Artificial intelligence (AI) is increasingly integrated into science education, yet the literature remains fragmented, lacking a unified framework that connects publication trends, curriculum implementation, and ethical considerations. Methods: A mixed-method systematic review was conducted that combined bibliometric analysis with qualitative content analysis. A total of 80 peer-reviewed studies were identified through Web of Science and Scopus databases (1990–2026) and analyzed using a structured coding scheme encompassing educational level, AI type, curriculum integration, ethical considerations, and pedagogical use. Results: Bibliometric analysis revealed an acceleration in publication growth since 2020, concentrated in higher education contexts. Qualitative analysis identified institutional and program-level integration as dominant (41%), while K-12 settings and teacher education remained underrepresented. Personalization and adaptive learning emerged as the main pedagogical application (59%), with transparency and student agency as the leading ethical concerns. Privacy, governance, and equity dimensions were critically underrepresented. Conclusion: The findings reveal structural imbalances and ethical gaps in the current integration of AI in science education. A multidimensional analytical framework is proposed, capturing the interdependencies between technological, pedagogical, and ethical dimensions. Future research should prioritize K-12 contexts, data governance, and systemic policy frameworks to support responsible AI integration.