Most people use four words as if they mean the same thing: AI, machine learning, generative AI, and ChatGPT. They are not the same. They are four boxes, and each one fits inside the one before it — which is why "we're adding AI" is almost never a useful sentence on its own.
Artificial Intelligence is the whole field — any machine doing something we would call thinking. It has existed since the 1950s. A chess program is AI.
Machine learning is one way of doing AI: instead of being given the rules, the machine works them out from examples. Spam filters and video recommendations are machine learning.
Generative AI is machine learning that makes new things — text, pictures, music — rather than only sorting things that already exist.
A large language model (LLM) is generative AI for text. ChatGPT, Claude and Gemini are all LLMs. This is the one almost everybody actually means when they say "AI".
When a company says it has "added AI", the useful question is: which of the four? A spam filter and a chatbot are both truthfully AI, and they are nothing alike.
It also tells you what to expect when it fails. A chess program does not invent moves that do not exist. A language model absolutely will invent a book that does not exist, because making new text is the whole thing it does.
It is not a search engine, and it is not a database. There is nothing inside it to look up.
It read an enormous amount of text and kept only the patterns — a bit like how you know how to write a birthday message without remembering every birthday message you have ever read.
Four different sizes, not four names for one thing. Every LLM is AI; almost nothing that is AI is an LLM.
The first activity in our free AI course asks you to nest these four boxes yourself, then explains what you just built. No sign-up.
▶ Play Lemonade Empire free