StatusIn preparation · the full piece will follow our opening
Confident isn’t correct
AI tools write fluent, certain-sounding answers. They can still be wrong, out of date or one-sided. A child who treats every answer as fact is handing over their judgement.
Practising disagreement
We ask learners to challenge an AI answer: find a weakness, ask for the other side, check a claim against a reliable source. Then they explain, in their own words, where they agree and where they don’t.
Fluent answers can still be wrong.
Why it builds thinking
Arguing with a tool is safe practice for arguing with ideas. It builds the habits of questioning, checking evidence and holding your ground for good reasons, which matter just as much with people as with machines.
What you’ll take away
- 01Fluent answers can still be wrong.
- 02Ask for the other side, then check.
- 03Questioning a tool is practice for questioning ideas.
