Exploring the integration of AI into Physics Education: Leveraging ChatGPT for Problem Generation

The rapid advancement of large language models (LLMs) presents an opportunity for educators to find innovative ways to include Artificial Intelligence (AI) into physics course design. Engaging with LLMs can support physics instructors in streamlining problem creation, enhancing diversity in problem sets, and expanding practice problems for student learning, including less explored physics topics. Our poster introduces two methodologies using ChatGPT for generating physics problems. The first approach leverages ChatGPT's generative abilities, to align problem generation with established problem styles by instructing the model to emulate contexts from question banks. The second approach involves conversational interaction, guiding iterative problem creation by segmenting tasks. Both approaches highlight techniques such as providing reference texts and creating subtasks to effectively guide the model. However, both approaches require careful human verification as the model tends to produce convincing yet flawed solutions, especially in unfamiliar problem domains. Drawing on our Fall 2023 classical mechanics course, we discuss the strengths, limitations and specific suggestions for integrating these approaches into physics course designs, offering insights for fellow educators.