2026 ASEE Annual Conference & Exposition

Engineering students often struggle to synthesize interdependent mechanics concepts—stress, strain, stiffness, and fatigue—into coherent mental models that guide design reasoning. This study examines how Generative AI dialogue can serve as a cognitive apprenticeship scaffold to promote conceptual learning in mechanics. The approach integrates the theoretical foundations of conceptual learning and cognitive apprenticeship to structure student–AI interaction around modeling, coaching, and reflective articulation.

Within a junior-level mechanical design course, ChatGPT was deployed as an adaptive conversational partner prompting students to reason through the relationships among stress distribution, deflection, fatigue endurance, and geometry. Students engaged in iterative AI-mediated questioning that mirrored expert reasoning sequences—linking analytical equations to physical interpretation and decision-making. Complementary JupyterLab notebooks provided computational grounding, allowing learners to test their evolving conceptual hypotheses through simulation and visualization.

Analysis of student reflections and in-class transcripts indicates that AI-facilitated dialogue enhanced metacognitive awareness, improved integration of multiple design parameters, and supported transfer of knowledge to unfamiliar design problems. The results suggest that generative AI can operationalize conceptual learning by making expert thought processes visible, personalized, and scalable. The presentation will discuss design of AI-mediated prompts, cognitive scaffolding strategies, and ethical considerations related to authorship, transparency, and student agency. This work advances understanding of how generative AI can function not merely as an information tool but as a pedagogical collaborator in conceptual engineering education.