In artificial intelligence, a [hallucination](https://en.m.wikipedia.org/wiki/Hallucination_(artificial_intelligence)) is a confident response by an artificial intelligence that does not seem to be justified by its training data.