Hallucination

A hallucination is output that is fluent, confident, and factually wrong. It happens because language models are trained to produce plausible continuations of text, not to verify claims — so when the training data is thin on a topic, the model fills the gap with something that reads correctly instead of declining to answer. Invented citations, fabricated product specifications, and plausible-sounding but non-existent API methods are all hallucinations. Grounding techniques such as retrieval-augmented generation reduce the rate substantially by giving the model verified source material to answer from, but no method eliminates it, which is why production systems keep a human in the loop for consequential decisions.