Just a scaffold, not an architect: AI as a literature mapping tool for students

Graduate students often struggle to grasp the full interdisciplinary scope of a research topic, a challenge potentially compounded by the rise of Generative AI and the unguided ways that students use these tools. In response, in a graduate writing course in the School of Computing and Information, I developed an “AI-assisted literature mapping” assignment to provide a structured, pedagogical use for GenAI. My approach seeks to break AI away from the role of an ersatz author (a drafter of plausible but deficient academic works of dubious academic integrity) and instead present GenAI to students as a “recommender engine” tool. Students use a single, highly-structured prompt that tasks the LLM to act as an interdisciplinary guide to students’ research questions, mapping related academic disciplines and generating potential library search queries for their topic. This small intervention nudges students to explore unfamiliar perspectives and engage more deeply with scholarly literature, both broadening their interdisciplinary horizons and directly scaffolding the library catalog exploration process. Students found the results of this activity to be genuinely exciting. In their reflections, my students showed a high degree of metacognitive engagement and intellectual discovery, expressing surprise and motivation to explore new research avenues they hadn't previously considered. This model offers a practical way to use GenAI to spark curiosity and build mature research habits, using the tool as a scaffold for discovery rather than an architect of the final product.