Hallucinations
Why AI sometimes sounds completely sure and is still wrong.
A hallucination is when an AI gives an answer that sounds confident and believable but is made up or wrong: a fake statistic, a book that doesn't exist, the wrong opening hours. It happens because the AI predicts likely-sounding words; it isn't checking facts.
How it works
- 01The AI predicts what a good answer would usually look like.
- 02If it doesn't actually know, it can still produce something that sounds right.
- 03Unless it searches the web or checks your documents, nothing stops the guess from reaching you.
The friend who never says "I don't know"
We all have that friend who answers every question with total confidence, even when they're guessing. They're often right, which is why you trust them, until they send you to a playground that closed two years ago. AI can be like that. Use it freely for ideas and drafts, and double-check anything where being wrong matters: health, money, legal questions and dates.
Where you'll use it
Health questions
Use AI to prepare questions for the pediatrician, not to replace their answer.
Facts in your content
Check statistics and quotes before they go into a post or a client proposal.
Opening hours and prices
Ask the AI to search the web, or check the official website yourself.
Your nap-time experiment
Teach the AI to flag its own guesses. Ask a factual question this way:
Tell me 5 facts about the history of advent calendars. For each fact, rate how confident you are (high, medium or low) and say whether I should double-check it. If you're not sure about something, say so instead of guessing.