How Generative AI Supports Scientific Inquiry and Science Practices in Science Education: A Scoping Review of Empirical Studies

Abstract

Generative artificial intelligence (GenAI) has rapidly entered science education, but its role in supporting scientific inquiry and science practices remains unclear. This scoping review mapped empirical studies on how GenAI has been used to support inquiry- and practice-oriented work in science education. Guided by JBI scoping review principles and PRISMA-ScR, the review used a reproducible open-data search route through OpenAlex and Crossref, supplemented by targeted manual discovery. Records published from 30 November 2022 onward were screened against predefined criteria, resulting in 14 included empirical journal articles. The reviewed studies clustered around three main pedagogical roles: learner-facing inquiry partners, teacher-facing design and co-design tools, and assessment or feedback agents. Investigation planning, explanation, and inquiry scaffolding were more common than data analysis, argumentation, and explicit epistemic reflection. Overall, the literature suggests that GenAI is currently used more often to structure, prompt, or evaluate inquiry than to support deeper disciplinary participation. Productive use depended heavily on teacher mediation, prompt quality, and learners' critical oversight. The review identifies both practical opportunities and persistent epistemic gaps for future research.