Artificial Intelligence–Assisted Science Learning in Elementary Education: A Systematic Literature Review on Pedagogical Models, Learning Outcomes, and Ethical Issues

Abstract

This systematic literature review examines the integration of Artificial Intelligence (AI) in elementary science education, focusing on pedagogical models, learning outcomes, and ethical considerations. Following the PRISMA 2020 framework, a comprehensive search of the Scopus database identified 637 initial records. After applying inclusion criteria (peer-reviewed empirical studies, elementary education context, AI-assisted science learning, published 2019–2025), 28 articles were selected for final analysis. Findings reveal three primary pedagogical approaches: inquiry-based learning with AI scaffolding, project-based learning enhanced by generative AI, and game-based learning for AI concept introduction. Documented learning outcomes include improved scientific literacy (N-gain scores up to 0.73), enhanced computational thinking skills, positive attitudes toward science, and development of science process skills. However, ethical concerns regarding algorithmic bias, data privacy, over-reliance on AI, and equitable access remain insufficiently addressed in current literature. This review synthesizes current evidence to propose a framework for responsible AI integration in elementary science education, identifying significant research gaps and providing recommendations for pedagogy, policy, and future research.