Course 1 · AI Foundations · Lesson 02
AI, ML & LLMs — The Map
Three letters everyone uses interchangeably — but they’re nested layers, not synonyms. AI is the field, ML the method, an LLM the engine, and the chatbot is the product. And underneath all of it: it calculates, it doesn’t think.
The one mental model
Three circles, nested. AI (the field) contains ML (the method), which contains the LLM (the engine). The chatbot you open is the product built around that engine. Different layers — not three names for the same thing.
Key terms
AI — the field
A goal, decades old: machines doing things that used to need a human. Already in spam filters, recommendations, auto-braking, map routing — not just chatbots.
Machine learning — the method
Instead of hand-writing rules (try writing the rules for “a cat”), you show it millions of examples and it learns the pattern. Learning from data is why AI got good.
LLM — the engine
A machine-learning model trained on an enormous amount of text, to predict what comes next. The thing humming under the chatbot.
Product, model & version
ChatGPT/Claude/Gemini are products (the car). Each runs a model family — Claude = Haiku · Sonnet · Opus — in numbered versions. Newer version = trained on more = smarter (Opus 4.8 > 4.5; GPT 5.5 > 5.4). Product ≠ model ≠ version.
The misconception to drop
✕
“The AI thinks.” It’s so fluent it feels like a mind weighing your question.
✓
It calculates. It predicts the most likely next piece of text from patterns it learned — no understanding, beliefs, or intent. Breathtaking pattern-matching, which is a very different thing from thinking. Knowing this is what makes you stop trusting blindly and start directing it.
Ask the AI Tutor
Pause anytime and ask — the tutor answers from this lesson’s material.
What’s the difference between an LLM and ChatGPT?
Is machine learning the same as AI?
Why do you say it doesn’t “think”?
Give me the whole map in one sentence.
Next lesson
03 — How an LLM Actually Works