Skip to content

AI glossary

What is an LLM?

An LLM, or large language model, is the kind of AI inside ChatGPT, Claude and Gemini.

By Published

Dani runs four apps with AI and has trained designers at PepsiCo.

The short answer
  • LLM stands for large language model.
  • It’s the engine inside ChatGPT, Claude and Gemini.
  • It predicts the most likely next words. That’s why it writes well, and why it can be wrong in a confident voice.

How it works, in three steps

You type a requestA question, an instruction, a document.
It predicts the next wordsOne after another, from everything it has read.
You get text that sounds rightUsually it is right. It doesn’t know when it isn’t.

When to trust it

Lean on it for

Sounding right is the job

  • First drafts and rewording
  • Summaries
  • Brainstorming
  • Explaining something simply

Check it on

It has to be true

  • Numbers
  • Dates and names
  • Money
  • Anything you’d forward to a client
From my own business

For a while my assistant wrote my morning summary and sent it to my phone.

Over nine days it arrived properly three times. Other mornings I got one word. “Morning.” Once just “Dani.”

When it did arrive, some numbers had been quietly rewritten. They looked right. They weren’t.

Now a plain script collects the numbers. The model writes only the words around them.

Words you’ll hear with it

WordWhat it means
ModelThe engine. Claude, GPT and Gemini are models.
TokenA piece of a word. Models read and write in tokens.
TrainingReading a huge amount of text to learn patterns. It happened before you arrived.
HallucinationA confident answer that isn’t true.
Context windowHow much it can read at once.

Three mistakes I see people make

  1. Treating it like a search engine.It predicts. It doesn’t look things up unless it has a search tool.
  2. Trusting a number because it’s written neatly.Neat is what it’s best at.
  3. Blaming the model.Often it was never given the facts.

A safer way to use one

Give it the factsPaste the document, the numbers, the message.
Ask for the wordingSummarize, draft, rewrite.
Check what must be trueNumbers, names, dates.
Want help setting this up in your business?

Talk it through with me. Tell me what your work looks like and we’ll find where AI helps first.

Talk it through with Danielle

Quick answers

Is an LLM the same as AI?

An LLM is one kind of AI, the kind that works with language. AI is the wider field, which also covers things like image recognition and recommendations.

What does LLM stand for?

Large language model. “Large” is its size, “language” is what it works in.

“Model” means a mathematical picture of how words follow each other. It is not a database of facts.

Is ChatGPT an LLM?

Almost. ChatGPT is the app. The LLM is the engine inside it.

Claude and Gemini work the same way: an app you talk to, with a model underneath.

Why does an LLM make things up?

Because it predicts likely words. It doesn’t look anything up unless it has a tool for that.

When the likely answer and the true answer differ, it still sounds sure. So give it the facts and let it do the wording.

What are examples of LLMs?

Claude from Anthropic, the GPT models from OpenAI that run ChatGPT, and Gemini from Google.

Does an LLM learn from my chats?

Not while you’re talking to it. The model doesn’t change during a chat. What it seems to remember comes from notes or memory features the app adds back in.

Next: how much it can read at once is its context window. One that takes steps on its own is an AI agent. Claude comes in more than one size.

AI Foundations starts here: give AI a clear picture of your business in a form you can reuse, so you stop explaining from zero.

Background: Wikipedia: Large language model