Large language model, explained
A large language model is a statistical model of language. During training it processes an enormous body of text and learns the patterns that connect words, facts and ideas. Given a prompt, it produces a response by predicting, one piece at a time, what text is most likely to follow. The products people use, such as ChatGPT, Gemini, Claude and Copilot, wrap one or more of these models in an interface, often with additional tools such as web search.
Two properties of LLMs matter for businesses. First, a model's built-in knowledge stops at its training data cutoff, and it reflects what the public web said about a brand at the time, accurate or not. Second, models can state things fluently that are not true. Assistants reduce both problems by retrieving current sources before answering and by showing citations.
For a brand, the practical question is not how the model works internally but what it has to work with: whether the sources it learned from and the sources it retrieves describe the brand clearly, consistently and credibly. That is the territory of LLM optimization and answer engine optimization.