Answers are grounded in this paper and cite their sources.
Artificial intelligence (AI) research in science and physics education has expanded rapidly, particularly since the emergence of generative AI. This study maps publication patterns, research characteristics, and citation connections across 14 selected science and physics education journals. OpenAlex searches using fifteen AI-related terms identified 966 unique records published between January 2015 and September 2026. AI-assisted screening retained 333 publications, which were coded using available titles and abstracts. Bibliometric analyses examined publication growth, thematic patterns, and citations within and between the two journal groups. Of the included publications, 87.1% appeared from 2023 onwards. Generative large language models and chatbots accounted for 58.9% of the corpus, while instruction and assessment together represented 48.3% of publications. Among 182 records classified as involving human participants, comparison groups were identifiable in 14.3%, effect sizes in 8.2%, and instrument validity evidence in 5.5%. These frequencies reflect information identifiable in bibliographic records, rather than the absence of these features in the full studies. Laboratory and experimental support was the principal educational function in 27 publications. Citation connections between the journal groups were limited: their reference sets had a Jaccard index of 0.047, and 15.6% of publications cited an included study from the other group. The findings document rapid growth and uneven research emphases within the selected venues. Independent human validation and full-text assessment are needed before drawing firm conclusions about methodological quality or educational effectiveness.
No takes yet. Share an insight, caveat, or question.
Konstantinos T. Kotsis (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: