A hallucination is when an AI chatbot gives an answer that sounds correct and confident but is actually wrong or made up.

A chatbot generates its reply by predicting likely patterns in language, even when it has searched the web or pulled in a document to help answer your question. That generating step is where things can go wrong. Every so often it produces a fact, date, quote, or source that simply doesn’t exist, or gets a detail wrong from something it did find. It states this with exactly the same confidence as something true, so there’s no obvious sign anything is wrong.

For parents, this matters most around homework and research. A chatbot can invent a book title, a historical event, or a statistic that was never real, and present it as plainly as anything correct. It’s worth teaching children to check anything that matters against another source, rather than trusting an answer just because it sounds certain.