Artificial Intelligence in Physics Courses to Support Active Learning Academic Article in Scopus uri icon

abstract

  • The integration of generative artificial intelligence (AI), particularly Large Language Models (LLMs) like OpenAI¿s ChatGPT and Microsoft¿s Copilot, is transforming educational methodologies, including undergraduate physics courses for engineering students. Despite their potential, these LLMs typically rely on statistical learning methods and often exhibit algebraic inaccuracies in solving standard university-level physics problems. This study explores the use of LLMs in physics courses for N = 91 freshman engineering students over two academic terms (Spring and Fall 2023). Students engaged in AI-assisted activities to solve physics problems and were asked to identify and correct the errors made by the chatbot. The outcomes were compared with those from traditional teaching methods without AI involvement, and no significant difference in student learning gains was found. To assess the impact of AI tools in education, a more detailed approach using pre-test and post-test instruments with control and experimental groups is necessary. Survey results revealed, however, that AI-assisted sessions enhanced student engagement, problem-solving skills, and understanding of physics concepts. Students also indicated a strong preference for AI-assisted activities, citing increased motivation and a firm belief in the educational benefits of using these tools. Our findings suggest that well-designed AI interventions can effectively complement traditional instructional methods, especially when the LLMs are integrated with symbolic computational tools like WolframAlpha to improve their accuracy. © 2024 Copyright held by the owner/author(s).

publication date

  • December 3, 2024