Level 0 · Curious
How does ChatGPT "think"?
It writes the most likely word, one after another.
ChatGPT doesn't look answers up in a database or "know" things the way you do. It learned from a huge amount of text, and what it does is calculate which word is most likely to come next.
It does that over and over, very fast, and that's how it builds full sentences. That's why it sounds so natural, and also why it sometimes makes things up with total confidence: a sentence can be likely without being true.
Try it today
Ask the AI for a very specific fact about your city, then ask: "Where did you get that?" See if it gives you a real source.
Test yourself
Why does AI sometimes make up facts?
- Because it lies on purpose
- Because it picks what's likely, not what's verified
- Because it's broken
See the answer
Because it picks what's likely, not what's verified. Its job is to produce likely text. If it doesn't have the fact, it still produces something that sounds right.
If you lead a team
If a team proposes using AI to answer customers, the key question is: what happens when it doesn't know the answer?