AI Is Already in Our Classrooms. The Question Is What We Teach Kids to Do With It.

The ABC's 7.30 took its cameras into Australian classrooms recently to see how schools are handling generative AI, and one comment from the report has stuck with me ever since. A UNSW lecturer put it plainly: it's not a question of whether students are ready for the AI revolution, because they're already in it. No waiting, no lead time, no phased rollout. The kids are living through it right now, and everything else in the debate has to start from that fact.

What the report showed was schools responding in genuinely different ways, and the differences are instructive. At Westbourne Grammar in Melbourne, students from Year 5 up are creating video games, generating art and getting AI feedback on their public speaking, and the principal was blunt that banning is never the solution. His focus is teaching students the safe, effective and ethical uses, and knowing when it's okay to augment their own thinking with the tools available. Meanwhile in NSW public schools, the department built its own tool, EduChat, and made a fascinating design choice. Ask it to write your essay and it politely refuses. Instead it asks you questions, nudging you to develop your own ideas, give it your own draft, do your own thinking.

I love that refusal, honestly, because it captures the whole argument in a single interaction. The problem was never the existence of AI in a classroom. The problem is AI that does the thinking so the child doesn't have to. Build the tool the other way, so it questions, prompts, scaffolds and pushes back, and suddenly you've got something closer to a tireless Socratic tutor than a cheating machine. One student in the report said she'd worried AI might erode her critical thinking, and found the tool was building it instead. And I haven't forgotten the assistant principal who pointed out that AI will never know when a child has missed breakfast or is quietly struggling. The humans stay at the centre. The tools just have to be designed, and taught, properly.

The risks named in the report are real ones. Over-reliance, weaker recall, hallucinated facts delivered with total confidence, and there's research suggesting students who lean on AI for writing show less of the brain engagement you'd want to see. I take all of that seriously. But every one of those risks is an argument for AI literacy, not AI avoidance. A child who knows what a hallucination is checks the output. A child who's never been taught believes it.

That is exactly the ground PrimarAI is built on. Primary-aged students learning what AI actually is, what it can do, where it goes wrong, and how to use it responsibly. Questioning outputs as a habit, recognising limitations, protecting personal information, and treating AI as a tool for learning rather than a substitute for it. And notice that even Westbourne starts at Year 5. I'd start earlier, because the habits of mind form in primary school, and if we wait until the kids have already picked up their habits from TikTok and a browser at home, we're remediating rather than teaching.

The technology will keep evolving, faster than any of us would like. Our responsibility is to make sure our children evolve with it, thoughtfully, safely, and with their own thinking firmly in the driver's seat.