Hallucination: definition

A hallucination is when a model states something false or invented as if it were fact, because it predicts plausible-sounding text rather than looking anything up.

Hallucination is the single biggest reason a demo that wows in the room fails in production. The model is not lying, it is filling a gap with the most likely words, which sometimes means a confident wrong answer.

You reduce it by grounding answers in retrieved data (RAG), asking the model to cite or say "I don’t know," and keeping a human in the loop on anything that carries real consequences. You never fully eliminate it, so you design for it.

Related terms

Retrieval-augmented generation (RAG): Retrieval-augmented generation is a technique that fetches relevant passages from your own documents at question time and hands them to the model, so its answer is grounded in your data instead of its training.
Large language model (LLM): A large language model is a neural network trained on huge amounts of text to predict the next token, which lets it write, summarise, classify, and answer questions across almost any topic.
Human in the loop: Human in the loop is a design where an AI system does the work but a person reviews, approves, or handles the cases the system is unsure about, keeping judgement and accountability with a human.