Not Punishment, Participation: Letting Students Co-Create AI Policy Recommendations in a Design Classroom

Why Read This from an HCI Perspective

This article reframes generative AI not as a question of whether it is a ‘good tool’ or a ‘bad tool,’ but as something to be understood through how people actually use it, where they get confused, and under what conditions they trust it—or hide its use. In particular, it matters that students are treated not as subjects to be regulated, but as co-designers of policy. For HCI and UX practitioners, it highlights that—more than the AI’s features—the way guidance is delivered, how exceptions are handled, and how responsibility is allocated are the key determinants of quality. For researchers, it suggests that participatory design can change not only the policy itself, but also user behavior.

CIT's Commentary

The core of this study is not about meticulously listing the ‘allowed boundaries of AI use,’ but about how those boundaries are understood—and by whom. It’s interesting that the moment students read the policy wording, they immediately think of score deductions and suspicion. That’s why interfaces that help users make judgments—such as task-specific examples and brief usage explanations—may be more important than surveillance-like mechanisms such as submitting chat logs. The format of a zine also functions as a safe intermediary: it’s not just an output, but a medium that helps surface experiences that are otherwise hard to articulate. Overall, the study shows that AI policy is ultimately an interaction design problem as well—one that must be designed together with transparency, pathways for intervention, and the potential for misunderstanding. Even in domestic universities or organizations like Naver and Kakao, it’s not enough to copy the same wording; you need to redesign the explanation approach and feedback loops to fit the local evaluation culture and power dynamics.

Questions to Consider While Reading

• Q.When students co-create policies, compliance may increase—but there’s also a risk that the rules become too loose. How can we design that balance?
• Q.Cutting AI use into numbers—like ‘allowed up to a certain percentage’—doesn’t reflect real work practices well. What interface or logging approach could be a better alternative?
• Q.When professors’ and students’ standards for AI use differ, trust breaks down easily. In an educational context, how much transparency is appropriate to ensure fairness?