Doan Winkel put the uncomfortable part plainly: if AI produces a response that would pass in your classroom, the question is too easy. Not just too easy for AI. Too easy, full stop. His follow-up series goes further, teaching students to build a panel of AI advisors and then argue with it, rather than copy from it.
The originals: Your Discussion Questions Are Too Easy for AI, plus the Board of Advisors series (Part 1, Part 2), all by Doan Winkel, How to Teach With AI.
The cheating panic is aiming at the wrong target
The common reaction to AI in the classroom is defensive: move everything in-class, go back to handwritten work, AI-proof the worksheet by making it lower-tech. That treats a design problem as a security problem. It also teaches students nothing about the tool they will use for the rest of their lives.
The better move is the harder one. Design questions that require judgment AI cannot fake, the ones that need a student's own reasoning, evidence, and defense. Then teach students to interrogate what AI gives them: to spot where it is confidently wrong, to push back, to make it justify itself. That is a more demanding assignment than anything a chatbot can quietly complete at midnight.
This is an assessment-design question
Underneath all of it is assessment design, which is the lane I spent my doctoral research in. The question "can AI do this assignment" is really the question "does this assignment measure thinking or measure compliance." AI is just the thing that finally made the difference impossible to ignore.
Here is where I land
Constructivist design changes the task, not the room. The usual reaction to AI is to move the same assignment into the classroom and watch students write it by hand. That does not fix anything. It just relocates the work. A constructivist assessment is built to create meaning for the student. The assignment itself drives the learning instead of sitting at the end of it as a box to check. When the work has a real reason to exist for that student, "can AI do it" stops being the question, because the point was never the output. The point was what the student built on the way there.
A redesign I would show a teacher this week. My final for my NIU graduate course is the living example. I built it expecting AI in the room. I broke one major assignment into eight parts, and students did the rough work in class workshops where they got feedback along the way. I suspect about four of my eighteen students used AI to polish the final, based on the formatting and on how well I know the tools. I called that good, and I meant it. They had already done the thinking in the workshops. Using AI to synthesize and polish work they reasoned through themselves is exactly the skill I want them to leave with. The other move I keep coming back to is reflection. This fall I am teaching eighth grade English language arts, and I am going to have students read a passage and reflect on what it actually means to them while we work on the mechanics of writing. AI cannot hand a student their own meaning.
Where teachers get this wrong. The most common bad reaction I see is the blanket ban. The sharpest version came from an educator I interviewed, who described a teacher who used AI to plan all of their instruction and then forbade students from touching it. That does not hold. If AI is good enough to build your lessons, it is good enough for students to learn to use well, and the honest move is to model that openly, which is why I put an AI disclosure slide in my keynote. Stoplight systems can help, green for open use, yellow for use it as a partner, red for hands off, but students will still cut corners where they feel they need to. The bigger issue underneath all of it is equity. Most districts do not give students any AI at all. When we do not supply it, students fall back on whatever they have at home: a phone, a family computer, or nothing. That gap is the inequity. If we are going to expect students to work with AI, we owe them equal access to it first.
If your teams are stuck reacting to AI instead of redesigning around it, that shift is something I help schools work through.
Dr. Chris Sanzeri is the founder of Evalve Consulting, an AI implementation practice for education organizations. He spent 15+ years in education leadership and builds custom AI tools, automations, and local AI systems for schools and districts.