To Police or to Guide: How Higher Education Computer Science Instructors Design and Implement Generative AI Policies
Based on interviews with 13 US computer science instructors, this study reveals that current generative AI policies often prioritize "AI-proofing" assessments over fostering student learning, leading to increased policing burdens and strained instructor-student relationships, and consequently recommends shifting toward learning-oriented policies that guide healthier AI usage.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are walking into a giant, bustling library where everyone is trying to learn how to build complex machines. For years, the rule was simple: you had to build the machine yourself, with your own hands, to prove you knew how it worked. But suddenly, a magical, super-fast robot assistant appeared. This robot can build the entire machine for you in seconds. Now, the teachers are in a bit of a panic. They are asking themselves: "If the robot does the work, how do I know the student actually learned anything? And if I ban the robot, will the student just sneak it in anyway?" This is the story of Computer Science Education in the age of Generative AI. Generative AI is a type of computer program that can create new things—like writing code, solving math problems, or explaining complex ideas—just by reading what you ask it. The big question isn't just about the robot; it's about the relationship between the teacher and the student. When a tool changes how we learn, does the teacher become a strict security guard trying to catch cheaters, or a helpful coach trying to guide the student through the maze?
This paper dives into that exact question by talking to 13 computer science instructors from universities across the United States. The researchers wanted to know: How are these teachers changing their rules (policies) about AI? Are they trying to stop students from using it, or are they trying to teach students how to use it wisely?
The study found that most teachers are currently acting like overworked security guards. They are mostly worried about "Assessment Harm," which is the fear that they can no longer tell if a student actually learned the material or just asked the robot to do it. Because they can't easily catch students using AI (the robot is too good at hiding its tracks), many teachers have switched to old-school tactics. They are banning laptops, switching to handwritten paper exams, and making sure students can't use any tools during tests. It's like a teacher deciding that since they can't stop students from bringing a cheat sheet into the room, they will just lock the doors, turn off the lights, and make everyone take a test in the dark with a pencil. While this might stop the cheating for a moment, the teachers admit it creates a tense, stressful atmosphere. It turns the classroom into a battlefield where the teacher is the "AI Police," constantly looking for rule-breakers, which makes students feel suspicious and less likely to ask for help.
However, the paper suggests that this "police" approach might be missing the bigger picture. The teachers noticed that when students use AI too much, they aren't just cheating; they are actually hurting their own brains. It's like a student who uses a calculator for every single math problem and then forgets how to do basic addition. They get an "illusion of competence"—they think they know the material because the robot explained it to them, but they haven't actually built the mental muscles to solve problems on their own. They also stop talking to each other and the teacher, becoming isolated because they are afraid of getting caught or because the robot is doing all the social "help-seeking" for them.
The researchers found that a few teachers are trying a different path: The Guide. Instead of banning the robot, these teachers are being very transparent about how to use it. They are saying, "It's okay to use the AI, but here is how you use it to learn, not to cheat." They show students how to ask the AI questions that make them think, rather than just asking for the answer. They are also using more low-stakes practice tests (formative assessments) where students can fail safely and realize, "Oh wow, I don't actually know this yet," before it's too late. These teachers are trying to build a classroom culture where students feel safe admitting when they used AI, so the teacher can help them fix their understanding rather than punishing them.
The paper doesn't claim that the "Guide" approach is a perfect, proven solution yet. It's more of a suggestion based on what these teachers are trying and how they feel. The researchers argue that simply trying to "AI-proof" the tests (making them impossible to cheat on) is a losing battle that strains the relationship between teachers and students. Instead, they suggest that teachers should focus on policies that help students learn with the technology, rather than just trying to ban it. The bottom line is that if teachers stop playing the role of the strict police officer and start acting more like a coach, they might be able to help students navigate this new world without losing their love for learning.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.