Hallucination
환각(할루시네이션)
When an AI states something untrue as if it were fact, with full confidence.
Overview
An LLM does not look up facts to answer. It writes plausible text. So when you ask about something it does not know, it often does not say "I don't know." It makes up an answer that sounds likely.
Typical cases are inventing papers or court rulings that do not exist, or getting numbers and dates wrong. Newer models do this less, but it has not disappeared.
An everyday example
It is like asking about restaurants in a neighborhood you have never visited, and getting made-up names and addresses of places that do not exist.
Why it matters
If you trust an AI's answer and pass it on, you can spread wrong information. Always check numbers, quotations, and legal and health information against the original source.
Also calledAI hallucination