The Biggest Mistake Schools Make With EdTech

By Heather Clayton Staker, Harvard-educated researcher and co-founder of Guide School

Over the past decade, schools have spent billions of dollars on laptops, software, and other edtech. And if you're a parent like me, you've probably been disappointed — though not entirely surprised — by how little those dollars actually changed.

I want to argue that there's one enormous design mistake underneath most of that disappointment. It's the same mistake schools are now rushing to repeat with AI, only faster and at greater scale. And once you see it, I don't think you'll look at laptops, AI, or the role of the teacher the same way again.

A quick note on where I'm coming from: I'm not a credentialed classroom teacher. I've spent years as a researcher and school builder, and for the past few months I've been a guide myself at our microschool, Zero School, in Indonesia. So some of what follows comes from data and observation across many Guide-model schools — and some of it comes from my own mistakes this year, standing in the room, resisting the urge to teach.

The horse and the engine

Imagine Henry Ford rolls the first Model T into town. Everyone gathers around, because they've never seen anything like it. One person studies it carefully and says, "This engine is incredible — let's bolt it onto a horse." So they strap the engine to the horse, hitch the horse to the wagon, and then wonder why transportation hasn't really changed.

That's what most schools have done with education technology.

Over the past decade, districts launched one-to-one initiatives. Parents bought devices. IT departments bought devices. We even bought special backpacks so our kids could haul laptops back and forth. And yet if you walk into most classrooms today, surprisingly little has changed. The schedule is the same. The teacher's role is barely different from what I saw in middle school. Students still move through content in age-based batches, everyone learning mostly the same thing at the same time — only now there are laptops.

We made life heavier for the horse.

Why "edtech doesn't work" is a fair complaint

When people tell me education technology has disappointed them, I understand completely. In a lot of schools, the laptop has become a distraction machine, a cheating machine, a YouTube machine. If that's what devices are doing in your school, I wouldn't want them either. Honestly, I'd rather convert the whole place into a Waldorf school — put the laptops away, bake bread, whittle wood, garden, take notes by hand.

That frustration is real and it's valid. But it rests on a cause-and-effect error.

The problem was never the laptop. The problem is that we kept the horse and bolted the engine onto it.

The exciting thing about laptops, AI, and adaptive software was never that they make a conventional classroom slightly more efficient. The exciting thing is that they make an entirely different model possible — but only if we're willing to let go of some assumptions that have been embedded in school for generations. And that turns out to be much harder than it sounds.

The mini-lesson that broke the model

Here's how deep these assumptions run. It shows up in a mistake I've watched teachers make hundreds of times — and made myself.

Years ago, when I first started helping teachers design blended learning, I'd run workshops where they built a full day in the life of a student in what I now call the guide model: a morning launch to energize the group, then independent work on learning apps for core academics, then project time in the afternoon with the laptops put away, while the guide holds one-on-one check-ins.

I'd ask teachers to start by designing the morning launch — a game or an open-ended discussion, something to bring everyone together before the independent academic work.

And almost every single time, something fascinating happened: the teacher slipped a mini-lesson into the launch. Ten minutes. Sometimes more. They'd explain, very reasonably, that they just wanted to cover the basics first, or make sure everyone understood the content, or offer a little direct instruction before students went off on their own.

It sounded responsible. It sounded caring. And it quietly broke the entire model. The moment you gather everyone to learn the same content at the same time, personalization is gone — you've walked right back into the conventional classroom.

These teachers weren't sabotaging anything. They were trying to help. The mini-lesson came from a belief most of us absorbed across thirteen years of school and had reinforced in teacher training.

The instruction fallacy

I've started calling that belief the instruction fallacy: the assumption that instruction is the scarce resource — that learning happens primarily because a knowledgeable adult stands at the front of a room and explains things clearly.

For most of human history, that was basically true. If you wanted expertise, you needed a human expert. If you wanted a personalized explanation, you needed a tutor. If you wanted feedback, you needed a teacher.

That's no longer the world our students live in. Today they can get explanations on demand, immediate feedback, adaptive practice, and repeated review, all tailored to exactly where they're stuck, all at their own pace.

The disruption isn't that computers are becoming teachers. The disruption is that instruction is no longer scarce — and once instruction stops being scarce, the role of the adult in the room has to change.

The test results that stopped me cold

That idea got very real for me this year at Zero School.

I'll be honest: when our mid-year math results came back, I was nervous. As a guide, I hadn't been standing at the front delivering lessons. I wasn't teaching math or reading or science. I was running morning launches, helping students set goals, handing out goal trackers, coaching them through AI-powered learning software, nudging them to try another strategy when they got stuck, and taking them to football matches, snorkeling trips, and farm projects.

There were plenty of moments where I wondered: Am I doing enough? Should I be explaining more? Should I just teach the lesson myself? Those urges creep in because most of us were trained to believe that explaining equals helping.

Then the results came back. As a group, our students had learned 3.6 times faster than the assessment projected they would — projections built on data from millions of students. Kids who'd spent half a year in a guide model had outpaced their expected growth by more than three times.

I sat and stared at the numbers. Because what hit me was this: so many of the moments when I felt like I wasn't doing enough were actually moments when students were learning to do more themselves. The apps delivered instruction at close to the right level. The kids were building persistence, learning how to get unstuck, learning how to find answers. And because they weren't trapped in age-based pacing, they moved faster — which freed up their afternoons for the things that actually make childhood rich.

Why good teachers revert

Before anyone starts feeling defensive, let me be clear about something. I don't think teachers slide back into lecturing because they're change-resistant, and I certainly don't think it's because they don't care. It's the opposite.

When you've spent twenty years being rewarded for explaining things well, stepping back feels irresponsible. Picture it: you're entrusted with 25 students. You care about them. You watch a child struggle, you know exactly what they're doing wrong, and every instinct says go help. And for all of educational history, helping meant explaining. Helping meant stepping in.

That's what makes this moment so emotionally hard. We're asking teachers to reconsider one of their deepest beliefs about what it even means to be helpful. I feel that pull myself, still, after everything I've seen.

Marcus and three days of basic subtraction

Let me tell you about Marcus.

Marcus is a second grader at Zero School this year. Early on, I noticed he kept getting stuck on subtraction word problems in his math app — not mildly stuck, stuck. Frustrated every day. Discouraged every day. I was sure I knew the problem: English is not his first language, and I assumed he couldn't parse the word problems well enough to solve them. The diagnosis felt obvious.

Then I did something that went against my instincts. Instead of sitting beside him reading each problem aloud in English, I opened the app's menu and assigned him a cluster of the most basic subtraction lessons imaginable — four minus three, that basic. It felt almost silly; Marcus is a bright kid, surely he had this cold.

But he was curious about those tutorials. Something in him seemed to sense there was a gap. He spent three days on the foundations, went back to the word problems, and within minutes was earning perfect scores. No tears, no frustration. The whole problem simply dissolved.

If I'd been functioning as a conventional teacher, I almost certainly would have misdiagnosed him. I'd have treated it as a language issue, pulled him into a small group, re-explained the problems, simplified the wording. What Marcus actually needed was three days closing a foundational skill gap — and because he had a laptop full of tutorials right there, he could act on that the moment we spotted it.

Now imagine Marcus in a conventional second-grade classroom. That moment never happens. The class keeps moving. He stays confused, and eventually he starts telling himself a story: maybe I'm bad at math, maybe school isn't for me, maybe the other kids are just smarter. None of it true. He needed three days with subtraction. That's it.

The paradox school leaders miss

Marcus points to a paradox worth sitting with: the less time adults spend delivering instruction, the more time they have for the work only humans can do.

When I stopped trying to be Marcus's instructor, I had far more capacity to be his guide — more capacity to notice, to encourage, to ask questions, to diagnose, to help him build confidence and find the right path.

People assume the guide model is less human because there are laptops in the room. What I've seen is the reverse. When software handles instruction, adults are freed up for coaching, mentoring, motivating, relationship-building, goal-setting, emotional support, character, and self-governance. The work becomes more human, not less.

The redesign question

So here's the question I'd leave you with. If I walked into your school tomorrow, I'd ask:

Where are adults currently doing work that a well-built system could do more precisely, more patiently, and more repeatedly than any human can?

And where could those same adults spend more time encouraging, mentoring, coaching, challenging, and inspiring — the things no software will ever do?

Because that's the whole shift. Laptops, the internet, and AI are not new tools for the old model. They're an invitation to redesign the model itself.

Most schools will use AI to make conventional instruction marginally more efficient — faster lesson plans, faster grading, faster feedback. The old model, married to some AI. But a much smaller group of schools — maybe including yours — will use the same technology to fundamentally rethink how learning works. My prediction is that the gap between those two groups will widen every single year, because the schools that redesign around the real strengths of both humans and AI will produce dramatically better outcomes.

I can't unsee what I've seen in Indonesia. And once you see it clearly, it's very hard to go back to thinking about school the same way.

Want help redesigning the model instead of bolting AI onto it?

If building the academic side of this feels overwhelming, you're not alone — that's exactly why we built these:

Core Engine — the AI-powered learning apps, student goal trackers, progress systems, and implementation support to stand up a guide-model academic block without reinventing everything from scratch.

From Teacher to Guide — our course with Arizona State University and the Clayton Christensen Institute, teaching the five shifts that turn teachers into guides, built on the model behind 3.6x academic growth.

Foundational Guide Certification — our in-person retreat in Nashville for educators ready to make this shift alongside a supportive cohort.

Next time on the podcast: what a real morning launch looks like when it isn't a disguised mini-lesson — and how to design the rest of the day so the model actually holds. Follow us on YouTube at @guideschooling or the Guide Schooling podcast on your favorite platform.