2026 ASEE Annual Conference & Exposition

This evidence-based practice paper investigates how students use a course specific GPT in the areas of reflective writing and metacognition in a first-year mechanical engineering seminar course. Generative AI tools are easily accessible to students and have the potential to affect the ways they interact with course content in a variety of ways. These tools have the potential to enhance learning but also can be easily misused. We believe that one way to support productive use of generative AI tools is to guide students in effective use of generative AI within a course specific GPT. Building on prior work with first-year mechanical engineering students showed that 42% were already using generative AI and 75% planned to use it in the future. Additionally, 70% believe these tools can enhance learning when used appropriately. This study examines the results of a survey of students in a first-year, introduction to mechanical engineering seminar course (N=234) to characterize how students use the course GPT (e.g., clarifying expectations, brainstorming/refining ideas, feedback on clarity/tone/structure, prompts to deepen reflection, pointers to textbook sections). We further evaluate how useful students perceive the tool to be for reflection and learning, and whether the availability of a reflective GPT lowers temptation to cheat (e.g., using AI to produce “reflections”). Survey analysis assesses perceived improvements in reflection quality, temptation to use AI, and students’ preference of course-aligned prompts and guidance over general-purpose AI outputs for reflective tasks.