hallucination

Definition

A hallucination is an LLM response that sounds correct but is factually wrong: fabricated citations, non-existent library functions, invented product features, false attributions. Hallucinations are a structural artefact of next-token prediction, not a bug. Mitigations: ground answers via RAG, require citations to retrieved sources, prefer extraction over generation when accuracy matters.

When to use

See also

• RAG — Retrieval-Augmented Generation — fetch relevant documents at query time and feed them to the LLM as context.
• LLM — Large Language Model — a transformer-based model trained on internet-scale text to generate and reason.