Answers are grounded in this paper and cite their sources.
Generative artificial intelligence (GenAI) is increasingly being adopted in education, yet its application to physics remains challenging because of the discipline’s reliance on mathematical, symbolic, and multimodal reasoning. This paper presents a PRISMA-based systematic mapping review of 146 studies published between January 2023 and June 2026 on the use of GenAI in physics education. The review examines the research landscape, approaches for supporting students and teachers, and methods for benchmarking and improving GenAI systems. The findings reveal rapid growth in research activity with widespread but uneven geographic distribution, with most studies focusing on higher education and relying heavily on third-party GenAI services. Student-support research primarily addresses conversational tutoring and personalized guidance, while teacher-support studies focus on lesson planning and teaching material development, professional development and teacher perceptions. A substantial body of work also benchmarks GenAI performance on physics-related tasks or proposes methods for improving it. Overall, the reviewed literature demonstrates considerable potential for GenAI in physics education while highlighting the need for further research on multilingual and inclusive systems, specialized educational uses, privacy-preserving approaches, improved multimodal support, and integration with complementary educational technologies.
No takes yet. Share an insight, caveat, or question.
Athanasopoulou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: