Foundations

AI, machine learning, LLM: what’s the difference?

Three words used as if they mean the same thing, untangled in five minutes.

5 min readPlain English

Read it in

In one sentence

AI is the whole field, machine learning is the main way it’s done today (learning from examples rather than hand-written rules), and an LLM is one kind of machine learning model, trained on huge amounts of text.

Think of it like…

AI is cooking: the whole business of making food. Machine learning is baking, one method among several. An LLM is sourdough: one thing baking makes, and the one everyone’s talking about. All sourdough is baked, but plenty of cooking isn’t baking.

How it works

The three terms nest inside each other, like boxes in boxes. Each one is a narrower, more specific part of the one before.

1AIAny computer system doing something that seems to need intelligence: recognizing speech, recommending a product, playing chess, writing an email. It includes older systems built entirely from rules people wrote by hand.
2Machine learningInstead of writing the rules, you show the system lots of examples and it works out the patterns itself. Spam filters and “customers also bought” suggestions are classic cases.
3Deep learningMachine learning that uses neural networks: many layers of simple calculations, loosely inspired by the brain, that can pick up very complex patterns in images, sound and text.
4LLMA large language model: a deep learning model trained on huge amounts of text to predict the next word. It’s what ChatGPT, Claude, Gemini and Copilot are built on.

In everyday talk, “AI” now usually means a chat assistant built on an LLM. On a product label it can mean almost anything, so ask what the product actually does.

Good at

Writing, rewriting and summarizing text
Answering questions in everyday language
Pulling structure out of messy notes

Bad at

Guaranteeing that what it says is true
Giving the same answer every time
Knowing your business unless you tell it

Where you’ll see it at work

The spam folder. Machine learning, not an LLM: your inbox learned what junk looks like from examples, so nobody had to write a rule for every scam.
“Customers also bought”. An online shop’s suggestions come from patterns in past orders. That’s machine learning too, and it never writes a word.
The reply drafted in your inbox. An LLM: a plumbing firm’s assistant drafts an answer to a customer’s question, and a person checks it before it’s sent.

Myth vs reality

Myth: AI and chatbots are the same thing.

Reality: Chat assistants are the most visible kind. Much of the AI you already use, like spam filters, fraud checks on card payments and product suggestions, works quietly and never chats.