Exploring the use of prompts and AI-generated images to elicit explanations in engineering education
Learning structural engineering concepts often presents challenges due to their abstract nature and the limitations of traditional teaching methods. One way to assess and develop conceptual understanding is through students' ability to generate high-quality explanations. While visualization too...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2026 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/456945 |
| Acceso en línea: | https://hdl.handle.net/2117/456945 https://dx.doi.org/10.1002/cae.70173 |
| Access Level: | acceso embargado |
| Palabra clave: | AI-generated images Engineering education Explanation skills Structural engineering Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial |
| Sumario: | Learning structural engineering concepts often presents challenges due to their abstract nature and the limitations of traditional teaching methods. One way to assess and develop conceptual understanding is through students' ability to generate high-quality explanations. While visualization tools are known to support this process, generative AI offers a novel approach for creating customized visual representations that may deepen student learning. This study investigates the potential of using generative AI tools—specifically, textual prompts and AI-generated images—to elicit and improve undergraduate students' explanations of key structural engineering concepts. Guided by two research questions, the study explores how these tools influence explanation quality and how the process of generating and reflecting on AI-produced images supports changes in conceptual understanding. Conducted in a steel structures design course with 29 students, the study asked participants to write initial explanations, generate images using AI tools, and then revise their explanations. Explanation quality was assessed using the SOLO taxonomy. Results showed that six of ten students who completed all stages advanced to higher SOLO levels, indicating improved conceptual depth. Students with lower initial understanding demonstrated the greatest improvement, while those with stronger prior knowledge experienced limited gains. The findings suggest that AI-generated images, when combined with structured guidance and clear instructional prompts, can support students in bridging abstract engineering concepts with concrete representations. However, visual tools alone are insufficient. This study emphasizes the need for intentional pedagogical design when integrating generative AI in engineering education and highlights future research opportunities to extend these approaches across diverse STEM contexts. |
|---|