What is a large language model (LLM)? A plain-English guide

The one-minute answer

A large language model is a computer program that has read an enormous amount of text and learned the patterns of language. Give it some words and it predicts what should come next, one small piece at a time. Do that thousands of times a second and you get an answer, an email, a summary or a poem.

• Large refers to the size of the model (billions of adjustable settings, called parameters) and the amount of text it learned from.
• Language is what it works with: words, and increasingly images, audio and code.
• Model means it is a mathematical system, not a database of stored answers.

How does an LLM learn?

Training happens in two broad stages:

1. Pre-training. The model reads a vast collection of text and practices one task over and over: guessing the next word. Each wrong guess slightly adjusts its internal settings. Over time it picks up grammar, facts, writing styles and some reasoning patterns.
2. Fine-tuning. The model is then trained further on examples of helpful conversations and on feedback from people, so it follows instructions, answers politely and refuses harmful requests.

The result does not “look up” answers like a search engine. It generates new text based on the patterns it learned.

What LLMs are good at

• Summarizing long documents into key points. Try our AI Text Summarizer.
• Rewriting text in a clearer or more formal tone. Try the AI Paraphrasing Tool.
• Drafting emails, reports and social posts. Try the AI Email Writer.
• Explaining topics at the level you ask for.
• Translating between many languages.
• Writing and explaining code.

Where they go wrong

They can be confidently wrong. An LLM produces text that sounds right. If it does not have reliable information, it may still write a fluent, specific and false answer. This is called a hallucination.

Their knowledge has a cut-off date. A model only knows what was in its training data, unless the app also connects it to web search or your documents.

They can reflect bias that exists in the text they learned from.

Small changes in wording change the answer. How you ask matters a lot. Our guide on how to write better AI prompts shows simple ways to get better results.

How to use LLMs safely

• Check important facts (health, money, legal, news) against a trusted source.
• Do not paste secrets such as passwords, bank details or confidential company data into consumer chatbots.
• Be specific about what you want: the audience, length, tone and format.
• Ask for sources, then open them yourself to confirm they exist and say what the AI claims.

LLM vs search engine

Feature	Search engine	LLM chatbot
Gives you	Links to existing pages	A newly written answer
Best for	Finding a specific source or current information	Explaining, drafting, summarizing, brainstorming
Main risk	Low-quality pages	Confident but wrong answers

Many tools now combine both: the LLM searches the web and writes an answer with links. Even then, click the links to check.

The bottom line

LLMs are powerful writing and thinking assistants. Treat them like a fast, well-read helper who is sometimes wrong, and you will get real value from them.