large language model

A large language model, or LLM, is the kind of artificial intelligence behind assistants like ChatGPT, Claude and Gemini. It is trained on enormous amounts of text and learns the statistical patterns of language, which words and ideas tend to follow which.

From that training, an LLM can do a surprising range of things: answer questions, write and edit text, translate, summarise, and even produce working code. It does this by predicting, one token at a time, what should come next given everything so far.

It helps to remember what an LLM is not. It has no live connection to the truth and no genuine understanding; it generates plausible text, which is usually right but can be confidently wrong, a failure known as hallucination. Knowing this is the key to using these tools well.