Facilitating Transfer of Learning through Problem-Posing in Physics Using Generative AI

This project serves the national interest by improving the preparation of future scientists and engineers through the implementation of evidence-based teaching practices in a physics course, designed to improve students’ problem-solving skills. This Level 1 Engaged Student Learning project will contribute to NSF’s mission to build the nation’s next-generation workforce of scientists and engineers. The significance of this project lies in invention of a set of design principles for incorporating generative artificial intelligence to improve problem-solving skills for students in large-enrollment introductory undergraduate STEM courses. The importance of the project lies in its strengthening of educational infrastructure by creating a scalable suite of materials that can be adapted and integrated into undergraduate STEM courses. Furthermore, the project also plans to develop a model for professional development of educators to prepare students in responsible use of generative AI for the purpose of generating novel problems that would improve their success on academic assessments. The project advances knowledge at the intersection of physics education research and generative AI by developing and testing an empirically grounded approach to using generative AI for problem-posing as a strategy for improving transfer of learning and academic performance. The overarching goal of this project is to advance our understanding of strategies that integrate structured problem-posing experiences, supported by generative AI, into a large-enrollment calculus-based mechanics course for future engineers and scientists. Under this scope, the project plans to investigate conditions under which improved student learning occurs when students use generative AI to generate, critique, and refine practice problems. The project plans to study the effectiveness of an instructional routine aligned with an integrated conceptual model that combines contemporary perspectives on transfer of learning with a framework on problem characterization. The project operationalizes a model of problem-posing trajectories across context, representation, and complexity as pathways toward adaptive expertise; illustrating which forms of generative AI scaffolding, sequencing, and verification support productive problem-posing and transfer. Furthermore, the project creates validated measures for posed-problem quality, verification behaviors, and reflective monitoring of the use of generative AI in student learning. The NSF IUSE: EDU Program supports research and development projects to improve the effectiveness of STEM education for all students. Through the Engaged Student Learning track, the program supports the creation, exploration, and implementation of promising practices and tools. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.