what is a large language model

A large language model is a computer program that has read an enormous amount of text — far more than any person could in a lifetime — and learned the statistical patterns of how words follow one another. When you type something, it continues the text in a way that fits those patterns. That is the whole trick at heart: it is an extraordinarily skilled guesser of which word comes next.

The word large refers to size. The model holds billions of adjustable numbers, called parameters, and was trained on trillions of words. At that much scale, simple next-word prediction starts to look like reasoning, translation, and writing — abilities nobody programmed in directly. They emerged from prediction alone.

It also helps to remember what it is not. An LLM has no memory of you between separate conversations unless a system is built to give it one, no senses or body, and no guaranteed grasp of truth. It produces fluent, plausible text, which is not the same thing as correct text — and because it sounds so confident, it is easy to forget the difference.