Hallucination (AI stating plausible but false things)
Japanese ハルシネーション
An AI hallucination is when an AI gives an answer that sounds right but is false or made up. It is why a person should check AI text before it goes on a website.
A hallucination is when an AI states something false in a confident voice. It is like a student who never leaves an exam question blank and writes a believable answer instead, so the result looks right at first glance. Made-up studies, web addresses that do not exist and wrong details are typical examples.
Watch out Asking several chatbots the same question helps you spot mistakes, but it does not prevent hallucinations. Check prices, dates and anything about the law against official sources yourself before AI-written text goes on your website.
Source: Google Cloud, IBM, arXiv "Why Language Models Hallucinate"