Resist Over-Trust

It means believing an AI reply just because it SOUNDS good — smooth, confident, detailed, with perfect grammar — instead of checking whether it is actually true. Fluency is about style; it is not proof. A false fact can sound just as smooth and sure as a true one, so a polished answer deserves a check, not blind trust.

An AI builds answers by predicting likely-sounding words, not by knowing the truth. So it can produce a fluent, certain-sounding sentence whether the fact inside it is right or wrong — it can even invent a real-looking source or statistic. Sometimes a confident, made-up answer is called a 'hallucination'. That is why how sure it sounds tells you almost nothing about whether it is true.

Treat a confident answer as a first draft you still need to verify. When a fact matters, cross-check it against a reliable, independent source — a trusted book, a reputable site, or a real expert. If the AI names a source, confirm it actually exists. And remember the rule: the smoother and surer an answer sounds, the more carefully you check the facts.

No — the goal is healthy checking, not fear. When nothing factual is at stake, like a fun idea for a made-up story, there is no fact to verify. The skill is noticing WHEN an answer makes a real factual claim — a date, a number, a source, advice about health or safety — and checking those before relying on them.

It is a Stage 6 lesson for roughly ages 11–12 — the top of KAVI Academy's primary AI-literacy track, written to be secondary-ready. It uses a story, compare-and-sort games, and decision practice with light reading.