Plain English
Hallucination
When an AI states something false or made up, in the same confident tone it uses for facts.
An answer that sounds right but isn’t: an invented fact, figure, quote, source or policy. It happens because a model predicts plausible words rather than looking facts up, unless it has sources to work from.
Why it matters
It’s the main reason every fact, name and number needs checking before it leaves your business. Giving the AI your own documents reduces it but doesn’t remove it.
At work
Asked about your returns policy, an assistant invents a 60-day window you’ve never offered. Give it your actual policy to work from, and check before anything reaches a customer.
Think of it like a confident dinner guest who fills the gaps in a story with details that sound right. They’re not lying; they just don’t know what they don’t know.
Related terms
Where it appears