Your Students Are Already Using AI. Here's How to Use That Against Them (In a Good Way)

Here's something most teachers don't know: you don't have to be the suspicious adult in the room when it comes to AI. Your students are already suspicious. They just haven't been asked to think it through out loud.

A new RAND Corporation survey of more than 1,200 students from middle school through college, published in early 2026, found that 67 percent of students agreed with the statement "The more students use AI for their schoolwork, the more it will harm their critical thinking skills," a jump of more than 10 percentage points in under a year. And this is students talking about themselves, not about some other generation or some abstract future kid. They're saying: I use this thing, and I think it might be making me worse at thinking.

That's not a crisis to manage. That's a discussion waiting to happen.

The Setup Most Teachers Miss

The instinct in a lot of schools right now is to treat AI as a policy problem. Write the honor code. Detect the cheating. Build the AI-free assignment. Those conversations matter, but they're mostly defensive. They don't teach students anything about how to reason, evaluate, or make good decisions under uncertainty.

What the RAND data reveals is an opening that's much more interesting than any policy debate. Students are using AI more and worrying about it more at the same time. Between May and December of 2025, the percentage of middle school students using AI for homework rose from 30 to 46 percent, while high schoolers went from 49 to 60 percent. The worry about critical thinking rose in parallel. Students are not oblivious. They're conflicted.

That conflict is not a problem to fix. In philosophy education, that's called productive dissonance, and it's one of the best entry points for Socratic dialogue that exists. When someone holds two things at once, "I use this" and "I think this might be bad for me," they're already doing something philosophically interesting. They're just not doing it out loud yet, in a structured way, with good questions pushing them further.

Your job is to give that conflict a room.

Opening the Discussion

Before you say anything, ask.

A clean entry point for this discussion doesn't require a clever hook or a dramatic video. It just requires one honest question placed in front of your students at the start of class.

Try something like this:

"A recent study found that about two out of three middle and high school students believe using AI for schoolwork is making their critical thinking worse. Do you agree? And if you do agree, why do you keep using it?"

That last clause is where the real philosophy lives. "Why do you keep using it?" is not an accusation. It's a genuine question. It invites students to think about the gap between what they believe and what they do, which is the starting point for a conversation about rationality, habit, autonomy, and how we actually make decisions.

Give them a minute to write their initial answer before anyone speaks. Writing first slows the reflex to just agree with whatever sounds smart. It also means every student has something to bring to the conversation, not just the three who always talk first.

Where to Take the Conversation

Once students are in, here are some directions worth steering toward.

The belief-behavior gap

If students agree that AI might hurt their thinking but use it anyway, ask them why. This isn't rhetorical. You're after the actual reasons. Some students will say it saves time. Some will say everyone else is doing it. Some will say they'll stop relying on it after school or college. Each of those reasons is philosophically interesting in its own way, and each opens a different line of questioning.

The time-saving reason leads naturally to: What are you making time for? What's the tradeoff? The "everyone else is doing it" reason is practically a gift, because it puts social pressure and independent reasoning into direct conflict, which is exactly what you want students to notice. The "I'll stop later" reason gets at how we think about habit formation and cognitive development, and whether it actually works that way.

What do we mean by critical thinking?

At some point, ask students to define what they think they're losing. This is harder than it sounds. "Critical thinking" is a phrase students have heard hundreds of times, but most of them have never been asked to explain what it actually is. When they try, the conversation gets good fast. Some will say it's problem-solving. Some will say it's asking questions. Some will say it's figuring things out on your own. You can push on all of those.

If AI gives you the answer to a math problem step by step, and you follow the steps and understand them, did you think critically? What if you just copied the answer without understanding? What's the difference?

Now you're doing epistemology. You're asking what it means to actually know something versus just having access to it.

The outsourcing question

Ask students whether there are other things they've outsourced to technology that they used to do with their brains. Navigation is an obvious one. Spelling. Phone numbers. Ask them whether they think those things changed how they think, and whether AI is different. This question is not leading anywhere predetermined. Students who think outsourcing calculation to a calculator freed up their thinking for harder things are making a real argument. Students who think AI is categorically different from a calculator are also making a real argument. Let them make it.

What Students Are Actually Worried About

It's worth paying attention to a piece of the RAND data that doesn't get as much coverage. Female students expressed greater concern than male students, with 75 percent of girls saying AI harms critical thinking compared with 59 percent of boys. That gap is worth sitting with. It might reflect different relationships to academic performance, different social pressures around authenticity, or something else entirely. You could build an entire discussion around why the concern is distributed unevenly.

Also worth noting: only about one in three middle and high school students said their school had a schoolwide rule about AI use. Most students are navigating this without institutional guidance. They're making individual judgment calls in an environment where the norms are genuinely unclear. That's a real-world reasoning problem, not just a school policy problem. It's worth naming that explicitly with students: You're all being asked to make decisions about something where the adults haven't figured it out yet either. How do you make a good decision in that kind of situation?

That question is not rhetorical. It's one of the most important questions in practical reasoning, and your students are living it right now.

The Deeper Lesson

A Socratic discussion built on this data is valuable on its own terms. Students will think harder, argue better, and leave class with a sharper understanding of their own behavior. That's worth the hour.

But the deeper lesson has to do with what it means to think for yourself at all. If students spend enough time in this conversation, most of them will eventually arrive at a version of the same problem: they don't entirely trust AI, but they also don't entirely trust themselves, and the question of when to rely on a tool and when to work through something the hard way is not one that anyone has answered well for them.

That's the real opening. Not AI policy. Not cheating detection. The question of how you develop as a thinker, what conditions that development requires, and what it costs you to skip the hard parts.

Students are already asking that question, in a vague and unexamined way, every time they open a chatbot and feel a flicker of guilt about it. Your job is to give that question a structure and take it seriously.

They're ready for it. The data says so.

If you want a classroom-ready framework for running this kind of discussion, How to Run a Socratic Discussion That Doesn't Suck lays out the mechanics in detail.

Want to put the ideas into practice? Try a free session at thinkingengine.org — no setup required, works in any browser.