2026 ASEE Annual Conference & Exposition

Whether we acknowledge it or not, students are already using AI tools such as ChatGPT in their coursework. The question for engineering educators is no longer whether to respond, but how. Since there is no effective way to prevent their use, engineering educators should adapt by designing assignments that require students to use generative AI in disciplined and transparent ways. This paper argues that homework grounded in guided exploration through AI dialogue—where students are instructed to solve a problem, design an experiment, or survey a technical topic through structured interaction with a large language model (LLM)—offers a pragmatic and pedagogically sound way to align engineering instruction with real-world practice.

One practical mechanism for such assignments is to require students to submit a link to their AI session as one of their deliverables. Major LLM platforms already provide persistent shareable URLs, which allow instructors to verify that the interaction occurred and to confirm that no two students or teams submit the same link. While no authentication system is foolproof, forging a convincing AI transcript with a matching verifiable link is significantly harder than fabricating a handwritten problem set. To further reinforce individual accountability and metacognitive awareness, students can be required to annotate their submission with two short reflections: (1) Why they posed one key question in the way they did, and (2) One moment when the model’s output surprised, confused, or misled them and how they resolved it. These lightweight prompts discourage passive tool use and foster active critical engagement—an essential habit for engineers.

The motivational benefits of this approach are supported by two well-established strands of educational research. First, decades of studies on intelligent tutoring systems show that adaptive, conversational scaffolding can increase persistence, reduce frustration, and improve perceived competence. Second, within the framework of self-determination theory, structured AI interaction offers both autonomy (students can choose what to ask and how to iterate) and competence affirmation (LLMs provide immediate feedback without judgment), conditions known to increase intrinsic interest. Rather than short-circuiting problem-solving, AI can serve as a low-stakes cognitive rehearsal partner that invites students to "lean into" addressing issues they might otherwise avoid.

This paper presents a framework for integrating AI-mediated assignments into engineering curricula, including sample prompts, implementation guidelines, and grading models (completion-based, rubric-based, and reflection-driven variants) in various engineering disciplines. It concludes with recommendations for future empirical work measuring cognitive gains, motivational shifts, and differential impacts across student demographics. By shifting from prohibiting AI to harnessing it as a structured learning tool, engineering educators can better prepare students for the future of professional practice while strengthening—not weakening—the integrity of their assessment systems.