We're asking the wrong question about AI and cheating
You're (probably) not going to catch the cheating
There are teachers (and departments, and whole schools) I know working very hard right now to build policies and procedures around catching kids using AI to cheat. I understand it, we are meeting the moment, and the instinct to protect academic integrity is real and right.
I keep coming back to a bigger question, the one I think we actually need to be asking: What are we trying to have students learn? Are the assignments and projects we’re giving them meeting this moment, or are we still living in the past?
Catching AI-generated text is a short-term fix for a long-term shift. If a student can submit an AI response and it passes as their thinking, the assignment might be the problem. Plus as each day moves past, I fear it is going to get more difficult to even prove if a kid used AI to cheat.
The Question Behind the Question
When a teacher asks me “what tool should I use to scan for AI?” I try to gently redirect: “What would you know about the student’s learning if a scanner didn’t exist?”
That’s not a gotcha. It’s the design question. If the only evidence of learning is a final product that AI can replicate, we’ve built an assessment on a shaky foundation; by the way that was true before ChatGPT existed.
The teachers who are navigating this most confidently right now aren’t the ones with the strictest AI bans. They’re the ones who redesigned their assignments to ask for things AI can’t give: specific classroom moments, personal reasoning, iterative drafts with reflection, oral defense of written work.
The redirect isn’t “stop policing AI.” It’s “design assignments worth policing.”
A few places to start:
- Ask students to explain their thinking process, not just their conclusion
- Build in a handwritten or in-class component tied to the larger project
- Assign something that requires *this class, this week* as the raw material; a discussion that happened, a text you read together, a problem unique to your room
- Use AI as a teaching tool, not just a threat: have students audit an AI-generated response for accuracy, bias, or missing nuance
These aren’t workarounds. They’re better pedagogy. The moment is pushing us toward it faster than we might have chosen, but the direction is right.
Tools Built for This Moment
I built two AI Gems specifically for educators navigating these exact questions. Both are inside my AI Educator Gems Bundle at katiepage.ai.
Assignment Audit
Run any existing assignment through five AI-vulnerability questions before you hand it out. It flags where AI could do the work for students and suggests targeted redesigns. Takes about 10 minutes per assignment.
Teaching Strategies
Get differentiated, AI-informed instructional strategies for any lesson goal or student population. Built for real classroom use, not theoretical frameworks.
Both gems, $25. [Get the bundle at katiepage.ai, sold separately too]
From the Community
If this is hitting close to home, if you’re the one in your building trying to redirect this conversation and it feels like pushing water uphill, reply and tell me where you’re stuck. I read every message, and sometimes just naming the specific friction helps.
Have a great week,
Katie