1.5
The Mistake That Explains Everything
It has a name, and once you know it, you cannot be fooled the same way twice.
What you'll leave with
By the end of this lesson, you'll understand hallucinations (what they are, why they happen, and what to do about them) and you'll leave with the judgment to use AI well rather than either blindly trusting it or dismissing it.
Why this matters
There is a specific thing AI does that confuses almost every new user. Once you understand it, you will use AI differently, and better, for the rest of your life.
The idea
Remember the previous lesson: AI generates language based on patterns, not facts. It produces text that looks like a correct answer would look. The problem is that it cannot always tell the difference between generating a correct answer and generating a convincing-looking incorrect one.
This has a name: hallucination. It is an odd word for a simple idea. When AI hallucinates, it produces something false with the same fluency and confidence as something true.
What to know
Examples of how this shows up:
- AI inventing a book that does not exist, but citing it with a real-sounding author and publisher
- AI generating a plausible-sounding statistic that has no source
- AI describing a person's career history with details that are partly invented
- AI answering a legal or medical question with authoritative language and subtle errors
Why does this happen? Because AI does not know facts the way you look something up. It generates the most plausible continuation of your prompt.
Plausible is not the same as true.
The honest mental model: AI writes at the level of a knowledgeable, confident person who might be making something up. You cannot tell from the tone. You have to check.
This is not a reason to stop using AI. It is a reason to use it well.
The practical rule:
Use AI to draft, brainstorm, and organize.
Verify anything factual before you act on it or share it.
The more consequential the decision, the more careful you should be.
Example
Ask it for a precise statistic on a niche topic you know well, or for the publication details of a study on something specific. Then spend thirty seconds searching for what it gave you.
Sometimes everything checks out: the tools have gotten much better, and many now search the web before answering. But when a detail does not check out, notice this: it arrived sounding exactly like the details that did. That is the lesson. The failures have become rarer and subtler, which makes the checking habit more important, not less.
Try this now
Ask AI to recommend a book on a specific topic you care about. Then take one of the titles it gives you and search for it. Is it real? This is a classic hallucination test, and it will make the lesson concrete very quickly.
Save this
AI writes at the level of a knowledgeable, confident person who might be making something up. You cannot tell from the tone. You have to check.
Quiet takeaway
AI is a tool. Not a genius, not a threat, not magic. It is a fast, well-read, occasionally overconfident assistant for language tasks. It does some things remarkably well. It fails in specific, predictable ways. Now you know both.
Next
In Module 2, we learn the skill that makes everything else work: how to ask.
Download template
A worksheet to go along with this lesson