Eliciting Student Understanding in Structural Engineering Classrooms Using Text-to-Image Generative Models

In this work-in-progress, we explore the use of prompt engineering and image generation to elicit student understanding of different concepts discussed in structural engineering courses. In the context of image generation, prompt engineering refers to the process of providing a text or image prompt to a machine learning model to generate an image that meets certain criteria. The quality of the prompt can significantly impact the quality of the generated image. Hence, prompt engineering is an important part of the image generation process. By providing a well-crafted prompt that accurately describes the desired image, the generative model will more likely produce an image that meets the desired criteria. In the context of this study, students created prompts that were converted into images using Text-to-image Generative Models. When adequately prompted, these images are meant to visually describe their understanding of the concepts we have discussed during traditional lectures. The participants were presented with their own image as well as others' images to elicit both agreements or concerns at two levels: their understanding of the concepts and their understanding of the importance of prompting in the foreseeable future of AI-based systems. Hitherto, we have tried to use the generation of images using natural language for answering these research questions: i) for students, to which extent the crafting prompts may elicit thinking about the structural engineering phenomena? and ii) for instructors, to which extent these represent the level of understanding of the students on these topics?