Mapping the Integration, Impacts, and Determinants of Artificial Intelligence in Science Education: A Systematic Literature Review

Background: AI is increasingly examined as a tool for science-specific instruction, feedback, inquiry and teacher preparation, but research designs and claimed outcomes vary. Methods: This systematic literature review applied PRISMA 2020 reporting to a title-only Scopus search through Watase Uake on 5 September 2026. Two phrases retrieved 123 and 114 raw records, respectively (237 total). Searches were restricted to Q1 journal publications dated 2023–2026. The initial selection retained 49 articles; eight were removed in a subsequent science-education relevance audit. Two further Q1-indexed Scopus reports identified outside the Watase export were assessed but excluded as unrelated plant-science research. The final corpus comprised 41 articles, synthesized descriptively and thematically with study designs differentiated. Results: The literature describes AI as a scaffold for inquiry, dialogic science reasoning, multimodal representation, feedback, learning analytics and science-teacher preparation. Some primary studies report context-dependent benefits for measured learning outcomes, but findings cannot be generalized across heterogeneous designs. Conclusions: Appropriate science-specific pedagogy, human–AI interaction, teacher competence and ethical oversight are central considerations.