L

Learn AI

Track progress · learn offline

Open

Lesson 5 of 6

When not to trust it

6 min read

It sounds so sure. Polite, fluent, confident — even when it's completely wrong. The danger isn't that it makes mistakes; it's that it makes them in exactly the same calm voice it uses when it's right.

Confident isn't the same as correct

An assistant can make things up — a wrong date, a shop that closed years ago, a rule that was never true. This even has a name: a [hallucination](glossary://hallucination). It isn't lying; it's predicting words that sound right. And made-up facts arrive in the same confident tone as real ones, so you can't catch them by vibe.

The tone tells you nothing. A confident answer can still be wrong — so a quick check is what actually protects you.

Check before you act

You don't need to distrust everything — just build one small habit: before you act on something that matters, verify it. Ask where it got the fact, look it up on the real source, or ask again a different way. This quick double-check is called [verification](glossary://verification), and it takes seconds.

Be extra careful with anything high-stakes — health, money, legal, or fresh news. It can also be out of date, since it doesn't automatically know what happened this week. Confirm before you rely on it.

The shape of it

The assistant confidently says a shop is open till midnight, and you're about to drive over. What's the smart move?

Continue in the app

Take the whole AI for Everyday Life course — tracked

Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.