What is a language model (LLM)?

A language model, often called an "LLM" (short for "large language model"), is an AI system trained to understand and generate text. ChatGPT, Claude, and Google Gemini are all language models. They aren't intelligent in the human sense — they are very well trained at predicting which word logically follows the previous one, based on huge amounts of text they read during training.

That training involves reading enormous amounts of text — books, websites, articles — from which the model learns how language, facts, and reasoning typically fit together. Afterward, you can ask the model a question, and it builds an answer word by word that fits what it has learned.

That "building word by word" also explains a well-known weakness: sometimes a language model makes up something that sounds convincing but is factually wrong. This is called a hallucination (see our article about that). The model isn't consciously "lying" — it's simply calculating what sounds plausible, and that isn't always the same as what's true.

Large language models are the engine behind chatbots, writing assistants, translation tools, and increasingly AI agents that carry out tasks for you. Once you understand that a language model is essentially an advanced pattern-matcher — not an all-knowing oracle — you can use it as a handy tool, without blindly trusting every answer.