2026 ASEE Annual Conference & Exposition

The emergence of large language models (LLMs) has introduced new possibilities for reimagining instructional strategies in engineering education. This work-in-progress paper presents an AI-assisted teaching model implemented in a core computer engineering course to support metacognitive development, foster active learning, and improve conceptual understanding.

Students engage in structured pre-class, in-class, and post-class activities involving advanced LLMs. Before class, they interact with LLMs to answer conceptual prompts, explore key ideas, and generate initial explanations. Class sessions shift focus from content delivery to dialogue and problem-solving, building on student insights. After class, students continue using LLMs to articulate what they’ve learned, refine their understanding, and reflect on remaining questions. To close the loop, students submit their chat transcripts to the instructor for analysis.

These transcripts are reviewed to identify patterns in student thinking, misconceptions, and areas of confusion. Insights from this data are used to adjust lectures, improve content delivery, and provide individualized feedback. This completes a feedback cycle that links AI-supported learning with responsive instruction. Early observations suggest that this approach enhances student engagement, supports self-regulated learning, and strengthens students' ability to explain complex concepts.

This project contributes a replicable model for integrating conversational AI into engineering instruction. It also highlights a pathway for using student-AI interactions as formative assessment tools. Ongoing work will assess the impact on learning outcomes, explore scalability to other courses, and refine instructor analytics methods.