Leveraging Generative AI for simulation-based physics experiments: a new approach to virtual learning about the real world

he increasing prevalence and power of generative artificial intelligence (AI) tools, provides both significant opportunities and threats to our current educational practices. We present a novel application of AI in physics instruction: engaging students in prompting, refining, and validating AI-constructed simulations of physical phenomena. We present both the approach and demonstrate its promise with a study comparing three instructional approaches in a laboratory focused on electric potentials: (i) students using physical equipment, (ii) students using a prebuilt simulator, and (iii) students using AI to generate a simulation. We found significant group differences in performance on conceptual assessments of the laboratory content (η² = 0.359). Post-hoc analysis showed that students in both the AI-generated and prebuilt simulation conditions scored significantly higher on the conceptual assessments than students in the physical equipment condition. Students in these groups also reported more favorable perceptions of the learning experience. Finally, this preliminary study highlights opportunities for developing students' modeling skills through the processes of designing, refining, and validating AI-generated simulations.