A large language model (LLM) is the engine behind almost every AI tool you have heard of — ChatGPT, Claude, Gemini, Copilot. It is a program trained on an enormous quantity of text that, given some words, predicts the words most likely to come next. Stacked billions of times, that simple trick produces fluent answers, drafts, and summaries.
What an LLM is not is a database or a search engine. It does not "look up" facts; it generates plausible-sounding text. That distinction is the whole reason the rest of this glossary exists — it explains why models hallucinate, why grounding matters, and why a context window has limits.
Why it matters at your desk. If you are a lawyer or a writer, an LLM is a tireless first-drafter that still needs a human editor — it will produce a confident citation that does not exist as readily as a correct one. If you are a teacher, the same model that drafts a lesson plan can fabricate a historical date. Treat the output as a draft from a fast, well-read intern who never says "I'm not sure."
The practical decision is not whether you are using an LLM — most of the tools in this directory are built on one — but how much you verify before the output leaves your hands.