What is an LLM? Watch
An LLM, or large language model, is an AI trained on a huge amount of text so it can understand and write language. You talk to one every time you use ChatGPT (OpenAI), Gemini (Google), Claude (Anthropic) or Llama (Meta).
Its core skill sounds simple: given some text, it predicts the most likely next word. Do that again and again, very well, and you get essays, emails, code and answers.
How an LLM answers you, step by step Watch
- Prompt: you type a question, like "What is the capital of India?"
- Tokens: the text is cut into small pieces called tokens. In English, one token is about three quarters of a word.
- Transformer: a neural network uses "attention" to see which words matter most to each other.
- Next token: it gives every possible next token a score and picks a likely one, such as "New".
- Repeat: "New" is added, and it predicts again: "Delhi". That is why answers appear word by word.
Why is it called large? Watch
Because of its size. An LLM has billions of parameters: numbers that get adjusted during training, a bit like the strength of connections in a brain. GPT-3, released in 2020, had 175 billion parameters. It is trained on books, websites and code, using thousands of powerful chips for weeks.
How an LLM is trained Watch
- Pre-training: it reads a huge amount of text and learns patterns of language, facts and reasoning by predicting the next word.
- Fine-tuning: it is trained on examples of good answers, so it follows instructions.
- Human feedback: people rate answers, and the model learns which ones are helpful, honest and safe.
What LLMs are great at Watch
- Explaining a topic simply, in English, Hindi and many other languages.
- Writing and fixing emails, posts and reports.
- Summarising long documents into key points.
- Writing and explaining code.
- Brainstorming ideas, names and plans.
Where LLMs go wrong Watch
| Problem | What it means | What you do |
|---|---|---|
| Hallucination | It can invent facts or sources and sound sure | Check names, numbers and links |
| Old knowledge | It knows only what was in its training data, unless it can search the web | Ask for sources, or use web search |
| Maths slips | It predicts words, so long calculations can go wrong | Ask it to show steps, or use a calculator |
| Bias | It can repeat unfair patterns from its data | Question one-sided answers |
Try it yourself Watch
Explain what a large language model is to a 12-year-old in India,
with one cricket example. Use 5 short bullet points.Lesson learned Watch
An LLM is a super-fast next-word predictor trained on huge text. Brilliant for drafts and explanations, but always check important facts.
Frequently asked questions
What is the full form of LLM?
LLM stands for large language model: an AI trained on huge amounts of text to understand and write language.
Is ChatGPT an LLM?
Yes. ChatGPT is an app built on OpenAI's GPT large language models. Gemini, Claude and Llama are other LLMs.
Does an LLM understand like a human?
Not in the human sense. It learns patterns in language very well and predicts likely words, which can look like understanding, but it can still be confidently wrong.