For educators

Using AI to skip the thinking makes kids lazy.We agree.

So we’re not going to argue with you. Below are the worries we hear most from teachers and homeschooling parents, one at a time — what we’ve built against each one, and where we fall short.

Try it yourself — free Play a game a kid built

Ages 8+ · Real JavaScript · Any browser · No ads · COPPA-compliant

Why AI and homework go so badly together

We’re not setting this up to knock it down. This is exactly how it goes wrong, and it’s why the worry is the right one to have.

The shortcut wins

Homework is done when it’s handed in. Whatever gets there fastest wins, and right now that’s a prompt.

The mistakes are invisible

A confident wrong paragraph looks exactly like a right one. The kid can’t see the error, so nothing corrects them.

They can’t explain it

No decisions were made, so there’s nothing to defend, extend or build on next week.

It has to be hidden

When work is graded, using AI well and using AI to cheat look the same from outside. So nobody gets taught the difference.

Notice what all four have in common. The work is being marked, the kid didn’t choose it, and they can’t tell whether what came back is any good. A game a kid decided to make is none of those things — and that’s not a rule we enforce, it’s just what building a game is like.

The six things people ask us about.

One at a time, with the honest limits attached. If something here doesn’t hold up for your kids, we’d rather you find that out on this page than three weeks in.

The worry

It does the thinking for them

This is the big one, and it’s the one most teachers raise first — more than half say they see AI used as a way around the work rather than through it.

What happens here instead

Here, the thinking comes first, because it has to. A kid describes the game they want, reads the plan we draft from it and fixes what’s wrong, and only then does any code get built — one step at a time. Every stage needs an answer only they have: who the player is, what makes it hard, what should happen next. There’s no “just do it for me” button, and a kid who tries to coast ends up with a bland game they didn’t choose and can’t steer.

Being straight with you: The first version of the code is written by AI. We’re not going to pretend otherwise — what we’re claiming is that the decisions around it aren’t.

The worry

They believe whatever it tells them

Fair. Most kids have no idea how confidently these tools get things wrong, and only about a third know AI can’t reliably tell fact from fiction.

What happens here instead

The difference here is that the answer runs. They press play, and whatever the AI got wrong shows up immediately — the enemy never appears, the jump feels floaty, the level is over in four seconds. Checking the work isn’t a lesson we bolt on afterwards; it’s the only way the game ever gets finished. Kids end up sceptical of AI output because the game keeps proving they should be.

Being straight with you: That teaches them to doubt code, not to doubt facts. Spotting a made-up statistic or a fake citation is still your lesson to teach.

The worry

It takes away the struggle they need

The strongest version of this: young kids need friction, patience and the experience of being stuck, so keep AI away from them entirely for now. That’s an argument about protecting effort, and it deserves a real answer.

What happens here instead

The struggle is the activity. Something breaks every single session, and the kid is the one who has to notice it, describe it and work out what to try — there’s no way around that, because a broken game doesn’t get to be finished. And it holds attention instead of scattering it: one game, come back to it tomorrow, then next week. It’s hard, and kids stick with it because they want the thing at the end.

Being straight with you: It’s still screen time, and it’s no substitute for reading, writing or building something with their hands. If your line is no AI at all at this age, we’re not going to try to talk you out of it.

The worry

Nothing they make is really theirs

You see this with AI homework constantly — a kid hands in something they feel nothing about and can’t tell you what’s in it an hour later.

What happens here instead

Ask a kid about their game and you get the opposite problem: they won’t stop talking. They picked the character, set how hard it is, threw out the boss they didn’t like, and they can send you a link to play it. That happens because they decide first and the AI helps second — never the other way round.

Being straight with you: Ownership comes from the choices, not from typing every line. A kid who wanted to write the whole thing from scratch would be better served by a plain code editor and a good book.

The worry

You can’t tell what the kid actually did

The thing that makes AI homework so corrosive: teachers say it’s eroding trust and making it much harder to tell what a student really knows.

What happens here instead

Nothing here is handed in, so nothing has to be hidden — and the whole process stays on screen. The idea they wrote, the plan they approved, the steps in order, a short explanation of what each build changed, and every edit since. Ask them to walk you through one step, or to add something small while you watch. You’ll know inside five minutes, and it’s a much nicer conversation than an originality checker.

Being straight with you: There’s no gradebook, no rubric, no class view. This is you and one kid, looking at their screen together.

The worry

These tools aren’t safe for kids

Almost every kid is using AI now, a lot of them are telling it about their feelings, and most have never had an adult talk to them about any of it. That’s a reasonable thing to be worried about.

What happens here instead

PlaySprig isn’t a chatbot with a friendly face on it. It makes games and does nothing else — it can’t browse the web, can’t hold an open-ended conversation, and won’t play the role of a friend or a counsellor, because none of that exists here. Everything it says is filtered to stay age-appropriate, the code it writes is checked before it runs, and games run sealed off from the network and the rest of the computer. Kids under 13 sign up with a parent’s email, and the parent approves the account.

Being straight with you: What a kid types does get sent to our AI providers so the game can be built. They’re all named in our children’s privacy policy, along with everything we store and how long we keep it.

What a kid gets better at

Six things, and not one of them is touched by typing a prompt and accepting whatever comes back.

Saying what you mean

Working out what you actually want before you ask for it — the difference between a good prompt and a lucky one.

Reading it properly

Going through a plan somebody else wrote and fixing it while fixing it is still easy.

Making sense of code

Finding the line that controls the thing they want to change, in code they didn’t write.

Getting unstuck

Something breaks every session. Working out why is the habit that carries into everything else.

Judging what came back

Deciding whether the AI actually did what was asked, and saying exactly what’s wrong with it.

Finishing something

Sticking with a project until it’s good enough to show people — rarer, and much more motivating, than starting one.

What the first hour looks like

One sitting, one playable game, nothing to install. It works the same for a class, a co-op afternoon, a club, or one kid at the kitchen table.

Kids under 13 need a parent’s email to get started, so send that home beforehand rather than losing the first ten minutes to it.

First 10 minutes
Everyone picks a kind of game and a setting, then answers questions about it — who you play as, what you’re trying to do, what makes it hard. No code yet.
Next 10 minutes
They read the plan that comes back, change what they don’t like, and say go. A good moment to have them explain their plan to the person next to them.
The rest of the hour
The game gets built a step at a time, and it’s playable after each one. They tweak numbers, add an enemy, fix what feels wrong — and leave with a link that works in any browser.

What PlaySprig isn’t

You’ve had enough things sold to you. Here’s what we don’t have.

It’s not a curriculum

No lesson plans, no scope and sequence, no gradebook. It’s a project kids build — the teaching around it is yours.

There are no teacher accounts yet

No class list, no dashboard. Everyone signs up individually, and under-13s need a parent’s email, so sort that out before the session rather than during it.

It’s not a tutor or a companion

It won’t help with the essay, explain fractions, or chat to a kid about their day. It builds games. That’s the whole thing.

We can’t show you test scores

No classroom study, no standards alignment, no learning-gains data. What we can show you is the product and the games kids have made with it.

Safe enough for a room full of 10-year-olds.

Most of what frightens people about kids and AI comes from open-ended chatbots — the advice, the companionship, the conversations that go anywhere. None of that exists here.

Children’s privacy policy Read the FAQ

Games only. No web browsing, no open-ended chat, no playing the role of a friend or a counsellor.
Everything it says is filtered to stay age-appropriate, and the code it writes is checked before it runs.
Games run sealed off — no network access, no reaching into files, nothing left behind on the computer.
Kids under 13 use a username and a parent’s email, and the parent approves the account.
No ads, no third-party trackers, no profiles built on children, nothing sold.

Questions we get asked

Isn’t the AI still doing the work?

It writes the first version of the code — we won’t pretend otherwise. But the kid decides what the game is, fixes the plan before anything gets built, judges whether what came back is right, and drives every change after that. That’s roughly how a professional works with these tools. The version you’re worried about is a kid doing none of those four things, and that path doesn’t produce a game worth keeping.

Do they learn to code, or just to prompt?

Both, in that order. It’s real JavaScript, it stays on screen, and they can change any line of it. Most kids start by nudging a number to see what moves, then start asking why. The fair worry is that a kid who never writes anything themselves can’t check what the AI wrote — which is why each step is small enough to read, and why they end up reading and changing all of them.

We don’t allow AI at this age. Is this different?

It’s a different activity — making something with AI, rather than having AI produce work that’s being marked — but it is still AI, and if your line is that kids this age shouldn’t use it, we’d rather you keep your line than take our word for it. The honest test is an hour on the free plan. Watch how much of it is a kid deciding, reading and fixing things, and judge from that.

How would I know they actually learned something?

Sit with them and go through it: the idea they wrote, the plan they approved, the steps in order, every edit since. Then ask for two things — tell me what this step changed, and add something small while I watch. A kid who let the AI do everything can’t manage the second one, and you’ll know within five minutes.

How do I run this with a group?

Right now, as individual accounts: everyone signs up, and under-13s need a parent’s email, so send that home first. An hour is enough for most kids to get to a playable game they can share. If you’re running a class, club or co-op and need more than that — class lists, a teacher view, one bill — email hello@playsprig.com and tell us what would actually help. We’d rather build the real thing than guess.

What happens to a kid’s data?

No ads, no third-party trackers, no profiles built on kids, nothing sold. Under-13s never give us an email at all — they use a username, and a parent’s email for permission, and that parent can ask us to delete everything at any time. What a kid types goes to our AI providers so the game can be built; they’re named in our children’s privacy policy, with everything we keep and for how long.

Make one yourself and see.

The free plan is the whole thing. Spend an hour building a game — you’ll know inside ten minutes whether a kid could coast through it, and that’s worth more than anything we can tell you. Teaching a group? hello@playsprig.com.

Start free — no card The same argument, for parents