Why AI Makes Things Up, and How to Fact-Check It
One day, ChatGPT gives you a date, a quote, or a source that looks perfectly credible… and is completely wrong. It’s not a bug, it’s a known behavior: it’s called a “ hallucination”. Understanding why it happens means never getting fooled again.
Why AI makes things up
A generative AI doesn’t “know” things the way an encyclopedia does. It predicts the most likely next word, over and over. Most of the time, it lands on the right answer. But when the information is missing, it doesn’t say “I don’t know”: it fills the gap with something plausible. Hence wrong answers delivered with complete confidence.
Important: even the best models still get things wrong some of the time. Hallucination isn’t a flaw someone forgot to fix, it’s inherent to how the AI works.
When to be most careful
- Precise numbers, dates, and statistics.
- Quotes, sources, references (it can invent them out of thin air).
- Very recent topics or highly specialized ones.
- Legal, medical, financial matters: never act without verifying.
3 habits so you never get fooled
- Ask for sources, then check them yourself. A source that doesn’t open is a source that was made up.
- Cross-check important information with a regular web search before using it.
- Use AI as a first draft, never as the final authority. The decision stays with you.
The right mindset
AI is a brilliant assistant that sometimes gets things wrong. Keeping a critical eye isn’t being against AI, it’s knowing how to use it. That’s exactly the posture of manage it, don’t be replaced by it : the tool proposes, you approve.
To sharpen these habits (a full module is devoted to them), start with the free quiz, then our online courses.