Free courses / Prompt Engineering Basics / Lesson 5
Verify, then use
A good prompt does not make a fact true.
Nothing in this course makes an answer correct. It makes answers more useful and easier to check. Names, numbers, dates, quotations and links still need an independent source.
Ask the model to separate what it is confident about from what it is guessing, then check the guesses first. That is the most efficient verification you can do.
When you cannot check something, do not use it. A well-written unverified claim is still an unverified claim.
- Better prompts do not make facts true.
- Check the flagged guesses first.
- Cannot check it? Do not use it.
BEEP / THE MYTH-BUSTER
A beautifully constructed prompt does not make a single word of the answer true. I want that on a poster, ideally above every desk in the country.
Try it
Ask a model for three facts about a topic you know well, with confidence flags. Check all three and note whether the flags matched the errors.
Give me three facts about a topic I know well, each with a confidence rating and a source I can open. Then tell me which one you would check first if you were me.
Then paste it into whichever assistant you use — ChatGPT, Claude, Gemini or Copilot. Not sure which? Meet them all.
Progress is saved on this device.
Questions about “Verify, then use”Asked by readers, answered by us. Open to read them.
Open to load them.
Ask about this page and we will answer it here, where the next person can read it too.
Nobody else can answer you, and you cannot answer anybody else. This is not a forum — we would rather say that than run a room we cannot watch at three in the morning.
Your question goes up with the answer and without your name. Nothing identifies you, ever.
If we cannot answer it we will tell you why, and that one stays between us.