Should kids use AI at school? What new research guidance says, in plain English

Australia's national education research body has told schools to use AI sparingly, avoid it entirely in the first three years of primary school, and bring it in carefully from there. The worry isn't that AI is bad. It's that children who lean on it too early never build the knowledge they need to think for themselves. Wherever you live, the reasoning applies to your child too. Here's what the guidance says and what you can do with it this week.

What did AERO actually say?

On 14 October 2026, the ABC reported new guidance from the Australian Education Research Organisation (AERO), the research body funded jointly by the Commonwealth and the states. In short, AERO advised:

• Use AI sparingly in classrooms, because overuse could harm long-term learning.
• Avoid AI entirely in the first three years of primary school.
• After that, use it carefully, alongside proven teaching such as explicit instruction.
• Introduce it properly in high school, because students will meet it in further study.
• At home, parents should limit AI use and avoid giving children unrestricted access to general chatbots.

AERO's chief executive, Dr Jenny Donovan, named the risk "cognitive outsourcing". In her words, it's when you "rely on the tool to do the thinking for you".

Why does it matter? Performance isn't learning

I see this every week with adults, not just children. When someone asks a tool for the answer, they get a good answer and learn almost nothing. When they ask it to explain, check their work or quiz them, they learn.

The research is still young. AERO itself said the evidence on AI and learning outcomes is limited. But the direction is clear: what matters is who does the thinking.

What should parents do?

You don't need to ban anything. You need a few simple rules.

1. Keep the early years screen-light

For children in the first years of school, the guidance is plain: no AI. Reading, drawing, building with blocks and talking things through do the work that matters at that age.

2. Make AI a coach, not a ghostwriter

For older children, change the question they ask. Swap "write my paragraph" for "ask me three questions about my paragraph". Swap "solve this" for "give me a hint, not the answer".

3. Use it together

Sit with your child when they use a chatbot. Ask them to explain what it said, in their own words. If they can't, the tool did the thinking.

4. Watch for the crutch

A useful test: could they do the task tomorrow without the tool? If not, slow down.

What should teachers do?

• Design tasks the tool can't finish alone. Ask for reasoning, drafts and reflection, not just a final product.
• Prefer tools that guide. Some states have built tools that walk students through reasoning instead of handing over answers. That's the kind of AI the research favours.
• Teach the "why" of the rules. Students follow a rule better when they understand cognitive outsourcing than when they're just told "no AI".

How can you tell if a child is outsourcing their thinking?

It isn't always obvious. A few signs that the tool is doing the work:

• The writing sounds older or more polished than the child usually sounds.
• They finish quickly but can't tell you what the work says.
• They get stuck the moment the tool isn't available.
• Their questions to the tool are commands ("write", "solve") rather than questions ("why", "how").

None of these means a child did something wrong. They mean the habit needs a nudge.

Questions to ask your child's school

If you want to know how your school is applying the guidance, these questions get useful answers:

1. In which year levels do students use AI, and for what?
2. Which tools are used, and do they guide reasoning or give answers?
3. How do teachers check that work shows a student's own thinking?
4. What should we do at home to match what happens in class?

Schools are working this out now, so a friendly question often helps more than a complaint.

How do you talk to your child about it?

Children respond better to reasons than to rules. Try something like this:

"Your brain is like a muscle. When the computer does the hard part, your muscle doesn't get the workout. We'll use AI when it helps you think, and leave it off when it would think for you."

Then let them help set the family rules. Ask them when they think AI is fair to use and when it isn't. You'll often find their rules are stricter than yours, and rules they wrote are rules they keep. Revisit the conversation each term, because what's right for a nine-year-old is different from what's right for a fourteen-year-old.

What doesn't the guidance say?

It doesn't say AI is harmful in every case, and it doesn't ask families to remove technology from the house. It's a caution about timing and habits, based on research that is still early. Expect the advice to be refined as more evidence comes in.

The same rule works for adults

The adults I teach run into exactly the same trap. In our Academy, learners build real projects with AI, but every lesson asks them to predict, explain and check before the AI does the work. It's slower for a day and much faster after a month.

A small task for tonight

Pick one thing your child (or you) used AI for this week. Redo it together, with the AI only allowed to ask questions. Notice how much more sticks.

Takeaway

The guidance isn't anti-AI. It's pro-thinking. Keep AI out of the early years, make it a coach later, and judge success by what the learner can do without it.

Source: ABC News, 14 October 2026