AI in your kid's homework
The conversation that helps them learn, vs the rule that doesn't hold.
By the end of this lesson
After this lesson, you'll have one short conversation ready for your kid about AI homework — that helps them learn instead of either banning AI or handing the work over.
AI showed up in your kid's school without much warning. Banning it doesn't work — they have it on their phone. Pretending they're not using it doesn't either. The honest middle is the same line you held when they were younger: some tools are fine for thinking with, not for handing in.
This lesson is about how to have one short conversation that holds — without accusation, without surveillance, without trying to catch them with software that doesn't work.
The two failure modes
There are two ways AI homework goes wrong, and they look different from the outside.
1. Handing in unread. Your kid asks AI for the answer, copies it, submits it. They couldn't walk you through what they wrote if you asked. Sometimes the writing suddenly sounds unlike them — cleaner, smoother, more adult. That isn't proof. Careful writers, second-language writers, and kids who've genuinely improved write that way too. It's a reason to ask kindly: "Walk me through this — what did you say?"
2. Skipping the struggle. Your kid uses AI as the first move on every problem — before they've thought, before they've sat with hard for ten minutes. They get fluent answers without the friction that builds the actual skill. The bright kids do this most, because the AI's hint feels easier than thinking. Subjects where this matters most: maths, writing, languages.
The first one is dishonesty. The second one is quietly worse — because it leaves the report card looking fine.
Your daughter's history essay on the Treaty of Waitangi suddenly sounds like a 30-year-old academic. That's not evidence she cheated. It's a reason to sit with her and ask, kindly, what she thinks each paragraph is saying.
The line that holds at any age
You don't need to ban anything. You need one principle that's plainer than any school's policy:
AI is fine for thinking with, not for handing in.
What's thinking with?
- Brainstorming — getting unstuck, finding a starting point.
- Asking AI to explain something the teacher said three different ways.
- Practicing maths, languages, writing — AI as a patient tutor that never sighs.
- Translating a slang-heavy text from a friend.
What's handing in?
- Asking AI for the answer and copying it without reading.
- Submitting an essay you couldn't summarise to your mum.
- Treating AI as the author of work you're claiming as yours — without telling anyone.
The rule works for primary school. It works for high school. It'll work for university. Adapt the language; the principle holds.
The same line worked for calculators, Wikipedia, spell-check. Each one was "the end of education" when it arrived. Each one became a tool you used or didn't use, depending on whether the goal was learning or finishing. AI is the next entry on that list.
The conversation that does more than any rule
Sit with them for 15 minutes. Open ChatGPT (or whichever they're already using). Ask the same question two different ways and watch the answers change.
Then say:
"Both of these answers sound confident. Only one might be right. Sometimes neither is. That's why YOU still have to think — the AI isn't a source of truth, it's a fluent guess."
Then add the line:
"AI is fine for thinking with — brainstorming, getting unstuck, explaining hard things — but it's not for handing in. If you couldn't walk me through what you wrote, it's not your work yet."
Don't lecture. Don't print this out. Just have it once. They'll remember it longer than any rule you set.
Your nephew shows you a maths problem he's stuck on. Instead of asking AI for the answer, set up a tutor prompt: "Be a patient maths tutor. Never give the final answer. Ask leading questions. If I'm stuck, give one small hint at a time." Then watch him work it out. That's AI doing its best work in education.
Try this with your kid this week
Pick a school topic they're working on right now. Sit with them for 15 minutes. Open ChatGPT free.
- Ask: "Why did the Treaty of Waitangi happen?" (Or whatever they're studying.) Read the answer together.
- Then ask: "Now explain that to me as if I'm 8 years old." Watch the answer change.
- Then ask: "How do you know which one is right?"
Have the conversation that follows. That's the lesson. No rule will land harder than this 15-minute experiment.
Keep this
- Rule — AI is fine for thinking with, not for handing in.
- Phrase — "Did you read this end to end? Walk me through it."
- Don't — Don't try to catch them with AI-detectors. The detectors don't work — they accuse innocent kids constantly, especially second-language students.
Pop quiz, no marks
-
Your son hands in an essay you'd never have predicted he'd write. What's the first conversation, not the punishment?
Show answer
"Did you read this end to end? Walk me through what you said." If he can't, that's where to start. The goal is learning, not catching.
-
Why are AI-detectors a bad idea for catching cheating?
Show answer
They have huge false-positive rates, especially on second-language English speakers. A 2023 Stanford study found a leading detector flagged real essays by non-native English students as AI-written 61% of the time.
Want to go deeper?
Each of these topics on Plain AI explores one idea from this lesson in more detail: