Glossary
What is an AI hallucination?
When an AI states something false as if it were true — a plausible-sounding guess that fills a gap in what it actually knows. It isn't lying; it doesn't know it's wrong.
A language model doesn't look facts up — it predicts likely-sounding text. When it doesn't actually know something, it still produces the most plausible-looking words, which can be confidently wrong. That's a hallucination: not a lie (it has no idea it's wrong), just a gap filled with a good guess.
Certainly — see Marren & Kohl (2021), J. Applied X, 14(3), which found a 37% effect.
Fluent, specific, and formatted like a real citation — and entirely made up.
You cut it down by not asking the model to recall from memory: give it the source (RAG), ask for citations you can check, and treat anything factual as a draft until verified. The fix is workflow, not a better-sounding answer.
Related terms
Appears in these lessons
See it live
Learn to catch it before you trust it
Spot and fix hallucinations on real examples — free in the app.