Traditional Education Won’t Survive AI

Seventy educators at an international conference had just spent twelve minutes in dialogue with an AI to generate an essay and to surface what they already knew. What started with generic arguments about assessment, ended somewhere specific: particular students they had taught, classrooms they had run, the discoveries that had quietly changed how they work. We compared them with a second essay the same AI had produced in sixty seconds, with no conversation at all.

One teacher named the difference. "Essay A is professional. But Essay B is mine. It says things only I could know." Another said she was glad she had done the thinking instead of being told what to think.

Both essays were AI-assisted. The difference was whether the person used it to skip the thinking or to reach thinking she could not have reached alone.

That distinction is the whole game, and most AI training in education has it backwards.

I run workshops, speaking sessions, and advisory for organizations working through these decisions. Work with me.

We are training teachers to do the thing we warn students against

Districts are investing heavily right now to teach educators to use AI as a "thought partner" for planning lessons, designing assessments, and making instructional decisions. Some of that is useful. AI does save real time on newsletters, documentation, and the administrative weight teachers should never have been carrying. But when the pitch shifts from saving time to offloading the judgment, we are training teachers to do the exact thing we tell students not to do. And a teacher who has handed over her own thinking cannot credibly ask a student to hold onto theirs.

The instinct behind the training is to make teachers efficient users of the tool. That instinct is an error because what makes someone a good user of AI is not fluency with the tool but knowing what is irreducibly theirs: the particular kid in front of them, the read of a room, the judgment that thirteen years of practice built and no model has access to. A teacher who knows what only she knows uses AI as a provocation and keeps the thinking.

The Intelligence Suite is a set of open, privacy-first classroom tools built on these frameworks. Explore the tools.

What the research shows about who loses

The way people actually prompt AI shows the split forming in real time. Misiejuk and colleagues (2026) analyzed 281 prompts from university students working with AI across a full semester. The students who produced the strongest work brought context, specification, and their own framing to the exchange. The students who produced the weakest work delegated the interpretation to the AI, then argued with whatever it returned. The weak prompts were reactive. The strong ones were authored. And the gap did not close with practice. It widened. Reliance did not teach the weaker students to think better. It let them rehearse a low-effort habit until it set.

A second study reaches underneath the habit to the cost. Surveying 299 STEM students across five North American universities, researchers found that routine, trust-driven AI use was associated with sharp declines in three things: reflection, critical thinking, and the basic desire to understand why something works. Reflection took the hardest hit. The more a student trusted the tool, the less they monitored their own thinking, caught their own errors, or sat in their own confusion long enough to learn anything from it.

The students most eroded were not the strugglers but the ones we call tech-ready: the confident, the tech-enthusiastic, the ones certain they had mastered the tool. Confidence in the tool was itself the risk factor. Neither academic seniority nor prior experience offered any protection. The students best positioned to benefit were the ones who lost the most.

Why this is worse for the child than for the teacher

For an adult, what these studies describe is atrophy: a capacity you built and then let weaken through disuse. It is concerning, and it is recoverable. The teachers in that conference room already had the thinking. The dialogue only surfaced it. Remove the shortcut and the capacity comes back.

The students those teachers go home to are not in that position. They are not rebuilding a capacity. They are trying to build one for the first time, during the years the architecture is supposed to form, and a tool that does the thinking for them means the building never starts. Atrophy is losing a muscle you once had. Foreclosure is never growing it. The teacher's real job is to protect the second group from the second thing. She cannot do it if she has quietly signed herself up for the first.

What actually won't survive

This is why the transmission model of education will not survive AI, and should not. For a century we built schools to deliver content and to measure how much of it was delivered, because delivery was the thing we knew how to count. AI delivers content faster and cheaper than any teacher ever could so if transfer is the point, the machine wins outright and there is nothing left for a teacher to do.

The replacement is not more technology training or AI literacy for teachers. It is closer to the opposite. Teach educators to name what they bring that the machine cannot, and to design their teaching around it. Build professional development the way you would build a good lesson. Dialogue that puts the teacher back in the position of doing the thinking, so he remembers what it feels like and can create the conditions for a student to do the same.

The teachers who left that session didn’t magically become better at using AI in any technical sense. They had not learned a single new prompt. They just remember that their expertise was worth surfacing, that the thinking was theirs, and that a classroom is one of the last places a child gets to build the thing before someone hands them a tool that offers to skip it.

The machine will keep getting better at delivering the answer. The only question that was ever worth asking is what we want a child to be able to do when the machine is not in the room.