Abstract: Design science research (DSR) is taught in university courses and used by students for their final theses. For successfully learning DSR, it is important to learn to apply it to real-world problems. However, students not only need to learn the new DSR paradigm (meta-level) but also need to develop an understanding of the problem domain (content-level). In this paper, we focus on content-level support (CLS), proposing an illustrative tool to aid students when learning to develop a conceptual design with DSR (e.g., for a prototype). Following the DSR paradigm, we deductively identify students’ issues and use the scaffolding approach to develop design requirements (DRs) and design principles (DPs). To offer AI-generated scaffolding, we use the generative language model (GLM) “GPT-3.” We evaluate our illustrative design through 13 expert interviews. Our results show that providing students with CLS is perceived to be helpful, but the interaction with the student needs to be designed carefully to circumvent unintended usage patterns. We contribute DPs and an illustrative instantiation thereof toward a DSR tool support ecosystem. More broadly, we contribute to the understanding of how humans can be supported by AI to solve problems, an important challenge in human-AI collaboration research.
Keywords: Problem solving, Artificial intelligence, Scaffolding, Design science research (DSR), Generative language models, GPT-3
Download This Article: JISE2023v34n3pp279-292.pdf
Recommended Citation: Memmert, L., Tavanapour, N., & Bittner, E. (2023). Learning by Doing: Educators’ Perspective on an Illustrative Tool for AI-Generated Scaffolding for Students in Conceptualizing Design Science Research Studies. Journal of Information Systems Education, 34(3), 279-292.