Large language models (LLMs)
The text-predicting systems behind chat assistants: what they are and what they can do.
What is it?
A large language model is a computer program trained on a very large amount of text to predict which words are likely to come next. By repeating that prediction one piece at a time, it can answer questions, write, summarize, translate and help with code.
An LLM does not look facts up in a database unless it is connected to one. It produces text that fits the patterns it learned, which is why its answers usually sound fluent but can still be wrong. Many current models also accept images, audio or files, not just text.
Which models use it?
- OpenAI's GPT models power ChatGPT.
- Anthropic's Claude models power the Claude assistant.
- Google's Gemini models power the Gemini assistant.
- Meta's Llama models are released with downloadable weights.
- Mistral AI offers both open-weight and hosted models.
How to use it
- Pick an assistant or model that fits your task and your privacy needs.
- Describe what you want in plain words, including the goal and who the result is for.
- Give the model the facts or text it should work from instead of relying on its memory.
- Read the answer critically and ask follow-up questions to refine it.
- Check important facts, numbers and quotes against a trusted source.
Benefits
- Handles many language tasks with one tool: writing, summarizing, explaining and coding.
- Works in plain language, so no programming is needed to start.
- Fast first drafts that you can then edit and improve.
- Can adapt tone, length and format on request.
Limits and risks
- Can state wrong information confidently (often called hallucination).
- Knowledge stops at its training data unless it is connected to search or documents.
- May reflect biases present in its training text.
- Anything you type may be stored or processed by the provider, so avoid sharing sensitive data.
Prompt template
Copy it, replace the parts in [brackets] and paste it into any AI chat.
Explain [topic] to me as if I am [audience, e.g. a beginner / a manager / a student]. Start with a two-sentence summary, then give three key points with one everyday example each. End with two common misunderstandings about [topic] and why they are wrong. If you are not sure about something, say so clearly instead of guessing.
More in Foundations
- Transformers and attentionThe neural network design that most modern language, image and speech models are built on.
- Tokens and context windowsHow models measure text, and why there is a limit to how much they can read at once.
- Training vs inferenceThe difference between teaching a model and using it, and why it matters to you.
- Open-weight vs closed modelsModels you can download and run yourself vs models you use only through a company's service.
- Choosing the right model (cost, speed, quality)How to match a model to your task instead of always using the biggest one.