AI Hallucinations
AI models state false things as confidently as true ones, because they predict plausible text rather than looking anything up. Here is why, and how to catch it.
Real lesson card · Page 1 of 3
Why AI invents facts
AI hallucination
When a language model generates information that sounds authoritative but is fabricated — a citation that doesn’t exist, a quote never said, a wrong date stated with full confidence.Example
Asked for a source, a model may produce a real-looking author, title, journal, and page number for a paper that was never written.Myth
A hallucination means the AI is broken or lying.Reality
The model has no notion of truth. It predicts likely next words, so a fluent falsehood and a fluent fact are produced by the same process — and look identical from the inside.- WhenYou need a fact you can be held to — legal, medical, financial, or a citationVerify it against a primary sourceWhyThe model’s confidence is not evidence, and fluency is not accuracy.
- WhenYou’re brainstorming, drafting, or exploringLet it generate freelyWhyInvention is a feature when you’ll edit and check the result afterward.
Recall check from the same lesson
A confident, well-formatted answer from an AI is good evidence that the answer is correct.
Review the explanation
Answer: False. Confidence and formatting are produced the same way whether the content is true or invented, so they say nothing about accuracy.
One sitting · 20–30 minutes
A focused session on AI hallucinations
LearnBench starts from what you already know — skip what you have, master what you’re missing.
Start now