Teaching children to use AI without outsourcing thinking

The question we hear most from parents and school leaders is the same: should children be allowed to use AI at all? Banning it does not work, and ignoring it is worse. The families and schools getting this right are teaching judgement rather than writing rules.

Start with disclosure instead of prohibition. A child should be free to use an AI tool on a defined task, provided they can say how they used it. This turns a discipline problem into a teachable conversation.

Insist on a first attempt. The rule that works best in practice is simple: think first, then check. A student who writes their own answer before asking a model keeps ownership of the reasoning and uses the tool to test it.

Teach verification explicitly. These systems produce fluent, confident, wrong answers. The habit of checking a claim against a real source is now a core literacy, not an advanced skill.

Redesign what you assess. If a task can be completed entirely by a chatbot, the task was measuring the wrong thing. Shift weight to oral defence, live work, iteration logs and applied projects.

Used this way, AI becomes what a calculator became for arithmetic — a tool that removes the drudgery and raises the level of the question. Used the other way, it quietly removes the thinking altogether.