Artificial Intelligence and Generative AI in Primary Science Education: A Bibliometric Analysis and Systematic Literature Review
TL;DR
Findings indicate that AI research in primary science education is expanding toward generative, conversational, and co-creative applications while conventional AI approaches continue to coexist.
Abstract
Artificial intelligence (AI) has increasingly been integrated into education, while the emergence of generative artificial intelligence (GenAI) has expanded its potential roles in teaching and learning. However, evidence regarding AI and GenAI specifically in primary science education remains fragmented. This study aimed to examine the thematic development, pedagogical applications, empirical outcomes, and limitations of AI and GenAI in primary science education. A bibliometric analysis combined with a systematic literature review (SLR) was conducted using Scopus-indexed publications. The bibliometric analysis included 33 relevant publications and employed VOSviewer to examine keyword co-occurrence, thematic clusters, density, and research recency. For the SLR, 24 studies underwent full-text assessment, resulting in 10 studies meeting the inclusion criteria. The bibliometric analysis identified four interconnected thematic clusters centered on AI, primary education, science education, generative AI, ChatGPT, and AI-supported science teaching. The SLR showed that AI and GenAI were used as learning partners, content and instructional-design generators, assessment technologies, and components of broader pedagogical approaches. Reported outcomes included improvements in achievement, problem posing, motivation, scientific communication, and engagement, although findings were heterogeneous. Several studies also reported limitations related to sample size, intervention duration, contextual dependence, and the difficulty of isolating AI effects from other instructional components. These findings indicate that AI research in primary science education is expanding toward generative, conversational, and co-creative applications while conventional AI approaches continue to coexist. Effective integration requires pedagogical alignment, developmental appropriateness, and teacher mediation.