Three initial attempts at Re-imagining STEM education in the era of Generative AI

The rapid advancement of Generative AI technology poses both serious challenges and unprecedented opportunities for physics and STEM education. This talk delves into three of my recent early-stage efforts in using LLMs to redesign certain common practices in physics education. First, I used LLM's ability to quickly generate large numbers of isomorphic problems following simple human input, to transform traditional synchronized exams. Each problem on an exam can be drawn from a large isomorphic problem bank that can be openly accessible to students prior to the exam. Therefore, the reformed exams can be immune to the negative impact posed by content-sharing platforms such as Chegg, and could also allow more flexible exam schedule as well as multiple attempts. Preliminary findings indicate that giving students open access to isomorphic problem banks only have small affects on students' performance, suggesting a potential paradigm shift in exam administration. The second initiative explores the possibility of leveraging GPT-3.5 to provide personalized feedback to students' open responses to conceptual questions with "few-shot learning" technique. Our initial results highlighted GenAI's ability to generate feedback that are indistinguishable from or even superior to that of human instructors in some cases, which could potentially lead to a grading assistant that reduces grading time by more than 70% for instructors. Finally, the third study explores using GenAI to addresses the critical issue of misinformation, especially in the context of climate science. We tasked ChatGPT to generate synthesized climate misinformation for students in a "physics of climate change" course, which allowed students to practice reasoning with data against mainstream climate misinformation without the need to collect actual online narratives. Those initial attempts are just a few examples of the many possibilities enabled by the latest GenAI technology for reimagining physics and STEM education of the future.