Before the Screen: What AI Should Look Like in an Elementary Classroom

Holly Durham
6 days ago
7 min read

Sept 14, 2026

In my last post, I argued that the students who will eventually use AI best may be the ones who spend their earliest years doing many things that don't involve AI at all.

Reading real books. Writing with pencils. Talking with other people. Noticing things. Wondering. Struggling a little before someone, or something, supplies the answer.

But that raises an obvious question: If AI is going to be part of our students' world, when and how do we begin teaching them to use it?

I don't think the answer is to put more devices in elementary students' hands.

I also don't think the answer is to pretend AI doesn't exist until middle or high school.

There's a space between those two positions, and I think that's where elementary teachers have an important opportunity. We can protect the experiences young children need to develop as readers, writers, thinkers, and people while also beginning to show them what technology can do when we use it purposefully.

The key may be surprisingly simple: start without it.

Some Things Are Worth Keeping Old School

We've spent the past several decades adding technology to classrooms, often with good intentions. Devices increased access to information. Digital platforms made differentiation easier. Online programs promised immediate feedback and individualized practice.

But just because something can be done on a screen doesn't mean it should be.

The American Academy of Pediatrics' most recent guidance on children and digital media makes an important distinction. Not all screen use is the same. High-quality digital experiences, particularly when an adult is involved, can support learning. But heavier and more solitary digital use can displace the very experiences young children need: reading, play, movement, conversation, and interaction with other people. For school-age children, excessive digital media use is also associated with weaker attention control and lower academic achievement. That matters in an elementary classroom.

Young children need to turn the pages of a book and go back to find something they missed. They need to draw what they noticed. They need to form letters and words by hand. They need to talk through an idea that isn't quite clear yet. They need to listen to someone else's interpretation and realize it isn't the same as their own.

There is research behind keeping some of this work decidedly low-tech, too. Studies of young children learning letters have found advantages for handwriting over typing, including stronger letter and word learning. Earlier brain-imaging research found that handwriting letters activated areas associated with the developing reading network in ways typing and tracing did not.

This doesn't mean a Chromebook is inherently bad or that every digital reading experience is inferior to print. Research comparing print and digital reading is more complicated than that. It does suggest, however, that we should be careful about replacing physical experiences simply because a digital version is more convenient.

AI Doesn't Have to Mean More Screen Time

This is where I think our conversations about AI in elementary schools can go off track.

When we talk about "using AI in the classroom," we tend to picture students sitting in front of computers entering prompts into a chatbot. But some of the most useful applications of AI in an elementary classroom may never require a child to touch a device.

A teacher can use AI to translate a note for a family. She can ask it to help her think of another way to explain a concept a student isn't understanding. She can generate examples at different levels of complexity or quickly gather background information about something that has captured her students' attention.

Imagine a second grader arrives at school fascinated by an insect he found on the playground. His teacher doesn't need to hand him a device and tell him to ask AI what it is.

They can look at it. Draw it. Describe it. Talk about what they notice. Look through a book about insects. Generate some questions.

And then the teacher might bring AI into the conversation - "What kind of insect could this be based on what we've noticed?"

Now AI hasn't replaced observation. The observation gave the AI interaction a purpose. And the next question might be even more important: "How do we know if that's right?"

Now students need the book again. Or another source. Or an expert. They are beginning to learn something much more valuable than how to type a prompt - AI can give us information, but we are responsible for deciding what to do with it.

Let Curiosity Come Before Convenience

Young children are remarkably good at asking questions.

Why do dogs turn around before they lie down?

How do birds know where to go in the winter?

Why is the moon sometimes out during the day?

How did people build pyramids without machines?

Those questions are exactly where there is a role for AI.

Ideally a child could go to the library, speak to the librarian, find books on the subject and read them during sustained silent reading time. But the library day may be in two weeks or the librarian now handle technology check out and all the campus copies and is so overwhelmed she can barely manage a book check out. And sustained silent reading time may be all but done away with - a principal may not see the value in it to raise test scores. AI can help a teacher follow a child's curiosity in a more timely way.

But there's an important difference between helping a child investigate a question and immediately answering it. If every question is met with "Let's ask AI," we teach children that wondering is simply the brief period before a machine tells us something.

Instead, we first ask:

What do you think?

What have you noticed?

What do we already know?

What should we ask about?

Who might know the answer?

Those questions slow the process down and allow for critical thinking to happen BEFORE we go to a computer - students must first consider how do I break down what I need to know and how do I think strategically about where to look.

The goal isn't to see how quickly we can get children to an answer. The goal is to use critical thinking and questioning to help them become better thinkers. AI becomes a tool to support natural curiosity, not a replacement for it.

AI Can Help Us Hear What Students Know

One area where I think these tools have enormous potential is with multilingual learners. Language can sometimes hide knowledge. A student may understand a science concept, have a sophisticated response to a story, or know exactly what she wants to say but not yet have the English vocabulary to communicate it.

Translation tools can help us get closer to the thinking underneath the language. That doesn't mean we stop teaching English. It means we don't confuse English proficiency with intellectual ability.

A child might explain an idea in a home language first. A teacher might use AI-assisted translation to better understand the response. Students might compare how an idea is expressed in two languages, notice where a translation doesn't quite capture the original meaning, or work together to find the English words that communicate an idea most precisely.

Used that way, AI isn't doing the thinking for the child. It's helping the teacher gain access to thinking that was already there. And perhaps just as importantly, it allows a child's home language to function as an important resource rather than something that has to disappear before academic learning can begin.

The Adult Still Matters Most

There is another reason I don't think young children need independent access to AI. They need us.

UNESCO's guidance on generative AI in education calls for age-appropriate use, protection of children's privacy, and adult oversight. That makes sense not only from a safety standpoint but from a learning standpoint.

The interesting part of an AI interaction with an eight-year-old isn't necessarily what the AI says. It's what happens between the child and the teacher afterward. This is again where questioning and critical thinking come in. After learning something from Google’s AI generator, ask:

"Do you believe that?"

"That doesn't match what we read yesterday. What should we do?"

"Why do you think it gave us that answer?"

"Can we ask the question more clearly or have we gotten bad information?"

"Which words would help it understand what we actually mean?"

“Should we skip the AI summary at the top and look at some of the websites?”

That's AI literacy. And notice how much of it is also just plain literacy. Students are considering word choice, audience, accuracy, evidence, meaning, and clarity. And most importantly they are learning that language matters and that information should be questioned rather than simply received.

The Rule Is: Human First

I don't think elementary teachers need a complicated framework for deciding when technology belongs in a lesson. Maybe we simply need a habit of asking what should happen first.

Experience first - Look. Touch. Listen. Build. Draw. Play. Observe.

Think first -Wonder. Predict. Talk. Read. Write. Try.

People first - Ask someone. Listen to a classmate. Explain your thinking. Work through something together.

Then technology.

Use it when it allows us to do something worthwhile that would otherwise be difficult: cross a language barrier, pursue a question, access information, hear another perspective, create something new, or investigate an idea more deeply. And sometimes, after asking those questions, we may decide we don't need the technology at all. And that's an important lesson, too.

The Goal Isn't to Prepare Children for AI

AI will change many times. And it will change quickly.

The tools our first graders eventually use in high school probably don't exist yet. Teaching young children to become proficient with today's AI platforms isn't really prepping them for their future. Teaching them to pay attention, read closely, write, wonder, question, listen, communicate clearly, and recognize when something doesn't make sense absolutely will.

So perhaps our goal in elementary school isn't to prepare children for AI, tIt's to build children who will eventually be prepared to use AI well.

That means giving them plenty of time away from screens. It means filling classrooms with real books, pencils, paper, art, conversation, movement, questions, and other people. It means protecting the slow, sometimes inefficient work through which children learn to think.

And it also means not being afraid to open a new tool when that tool can help a child go somewhere worthwhile.

AI should expand a child's world. It should never be the world through which the child experiences everything.