2026 ASEE Annual Conference & Exposition

Introduction
Generative AI tools such as ChatGPT[1] are increasingly used by students in biomedical engineering (BME) courses, yet little is known about how these tools influence students’ reasoning when they are used explicitly for verification and critique rather than answer generation. While prior engineering education studies have examined student perceptions of ChatGPT and AI access policies, fewer have explored structured post-solution comparison between student reasoning and model output in analytically intensive courses.

This work-in-progress (WIP) study examines the integration of ChatGPT into an upper-level, computation-focused BME course (approximately 50 junior-level students), where students were required to compare their own analytical and computational solutions with AI-generated responses and reflect on discrepancies. To accommodate this integration, selected assignments were redesigned to reduce repetitive manual computation and instead emphasize model interpretation, error checking, and evaluation of solution strategies.

Initial implementation reflections suggest that students experienced ChatGPT differently depending on task type: the tool was generally perceived as useful for programming and debugging, but more confusing and occasionally misleading for analytical mathematics. In several instances, students reported that ChatGPT reinforced incorrect assumptions embedded in prompts, leading to unproductive exchanges and increased frustration during early stages of conceptual understanding. These observations underscore the importance of scaffolding AI use and explicitly teaching students how to evaluate, rather than defer to, model outputs.

Following full institutional review board approval, a broader anonymous survey has been deployed across the biomedical engineering student population to contextualize the single-course implementation within department-wide practices. Ongoing work will analyze these data to better understand how structured AI engagement shapes students’ reasoning, trust, and problem-solving strategies in engineering education.