Human-AI Co-Creation in Engineering Education: Leveraging ChatGPT-4 to Develop Authentic Case Studies

Abstract

While the integration of real-world decision-making scenarios into engineering education has proven to enrich experiential learning, developing such authentic case studies remains a persistent challenge. In this exploratory study, we investigate the potential of ChatGPT-4 as a co-producer in the creation of authentic engineering case studies. Rooted in interdisciplinary educational theories, our investigation is set within the context of the graduate-level course "Managing Big Data Projects". We introduce a structured framework that helps faculty engage with ChatGPT-4 through guided prompts and iterative conversations. We also conduct a quantitative linguistic assessment of the generated case study, using standard Natural Language Processing (NLP) metrics. The results revealed that while ChatGPT-4 succeeded to a certain extent in creating authentic case studies that stimulate problem solving and decision making, it showed several limitations such as lack of narrative depth, weak storytelling, occasional fake citations, limited memory capacity, readability constraints, and an inability to follow detailed instructions. Our findings underscore the importance of human oversight and "human-in-the-loop"approaches in leveraging generative AI for educational design.