L

Learn AI

Track progress · learn offline

Open

Lesson 9 of 9

Trust, then verify

6 min read

You've picked up real skill now — the window, the right model, sources, search, tools, memory. One habit ties them together and keeps you out of trouble: knowing when to double-check.

Confident is not the same as correct

An AI says everything in the same calm, fluent voice — the true bits and the made-up bits sound exactly alike. It has no built-in tell for "I'm sure" versus "I'm guessing." Every so often it will state something false as plainly as a fact; this is called a hallucination. The danger isn't that it's wrong sometimes — it's that wrong and right arrive in the same confident tone, so you can't spot the difference by ear.

An AI's confident tone tells you nothing about whether it's right. True and false come out sounding identical — so you can't judge accuracy from how sure it sounds.

Verify what matters

You don't need to fact-check everything — that would defeat the point. Match your caution to the stakes. A brainstorm or a rough draft? Trust it and move on. A medical dose, a legal clause, a figure going into a report, a name you'll quote publicly? Verify before you rely. And you already know the tools: have it search and cite a source, or run the numbers as code, or paste in the real document. A grounded answer you can check beats a confident one you can't.

Scale your checking to the stakes: trust the low-risk, verify the high-risk. Lean on the very tools from this course — search, code, your own sources — to turn an unverifiable claim into a checkable one.

A quick test before you rely on something: could you show where it's true — a source, a calculation, the original document? If not, and it matters, that's your cue to verify.

The gist

An AI gives you a specific medical dosage, stated very confidently. Before acting on it, what's the right move?

Continue in the app

Take the whole Working with AI course — tracked

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