Stop Policing for AI—Start Evaluating Citations Instead

Instead of trying to catch students using AI, maybe we should be teaching them how to cite their sources correctly?

There’s been no shortage of hot takes on how higher education should respond to AI in the classroom. Should we ban it? Embrace it? Build AI-proof assignments?

I agree with many of the concerns. Yes, students are using AI tools. Yes, it’s changing how they write. And yes, instructors need a response. But my approach from day one has been a little different—and honestly, much more effective.

Why We Shouldn’t Use AI Detectors

For starters, most of the schools I work with prohibit the use of AI detectors, and they have good reason to. Tools like GPTZero and Turnitin’s AI feature have been shown to be inaccurate and discriminatory—especially toward students on the autism spectrum.

That’s a major issue on its own, but even beyond that, AI detection creates a punitive and adversarial classroom dynamic. I’m not trying to catch students in a game of "gotcha." I’m trying to teach them how to write with integrity. There’s already too much policing in education, which we know prohibits learning.

What I Do Instead: Grade for Citations

Instead of policing AI use, I evaluate citation accuracy. Every rubric I use—regardless of course or assignment—includes a category for:

• In-text citations

• Reference list or Works Cited

• Proper formatting (APA or MLA)

And here's the key insight: when students use AI tools, they almost never cite. That’s an automatic deduction in my rubric. No in-text citation? No credit for citations. Just a pasted Reference at the end? Still partial credit. Incorrect formatting? More lost points.

The result is that students who rely on AI quickly realize that turning in work without credible sourcing hurts their grade—without me ever needing to prove they used AI at all.

Why This Works

Once students see how citation impacts their performance, they start to pay attention. Some try to backfill citations, but many do it incorrectly, which opens the door for real-time feedback and instruction. In the process, I’m teaching:

• How to track sources

• Why citation matters

• How to apply citation styles correctly

• That AI use isn’t inherently wrong—but it must be transparent

I even allow students to cite AI, using APA’s guidance on how to cite ChatGPT and other tools. But here's the interesting part: very few of them ever do. Whether it’s oversight, uncertainty, or stigma—I’m not sure.

But the outcome is clear: focusing on citations gives me a structured, educational way to respond to AI writing—without the unreliable tech, adversarial tone, or legal gray areas of AI detection software.

AI and the Future

AI isn’t going anywhere. But that doesn’t mean we need to become surveillance officers in our own classrooms. Instead of spending hours worrying over detection tools, let’s teach what matters: how to identify, use, and cite sources responsibly.

Let’s stop policing for AI and start grading like educators again.

If you're an instructor, how are you navigating AI in your classroom? Do you grade citations like this—or have another strategy that works?

Subscribe for more posts on teaching, writing, and academic life in the age of AI.