Skip to content

AI Hallucination

An AI hallucination is when a language model states something confidently that is factually wrong: a made-up citation, a wrong date, an invented quote, a fabricated statistic. All current models do this to varying degrees.

The mitigation isn’t a smarter model, it’s a better process: retrieval augmentation, human fact-checking on every claim, source citations that can be verified. AI drafts should never publish without a human eye on the specifics.

Related terms

See also on Miss Pepper AI