What is AI, anyway?

Programs that learn patterns from examples instead of following hand-coded rules. Today, when people say "AI", they almost always mean LLMs(Large Language Model).

You've been using AI for years

AI vs. ML vs. LLM

These get used interchangeably in the news. They're actually nested:

┌─ AI ──────────────────────────────────────────┐
│  Machines doing things that needed             │
│  human intelligence.                            │
│   ┌─ Machine Learning (ML) ──────────────┐    │
│   │  AI that learns from data             │    │
│   │  instead of hand-coded rules.         │    │
│   │   ┌─ Large Language Models (LLMs) ─┐ │    │
│   │   │  ML trained on huge amounts     │ │    │
│   │   │  of text. ChatGPT, Claude, etc. │ │    │
│   │   └────────────────────────────────────┘ │    │
│   └────────────────────────────────────────────┘    │
└──────────────────────────────────────────────────┘

When someone says "AI" in 2026, they usually mean LLMs. For most of this course, "AI" = "LLMs".

How AI learns — what "machine learning" actually means

Old-school software was rules. A 1995 spam filter was an engineer writing if email contains "free $$$" then spam. Brittle, exhausting, and useless against patterns nobody thought to hand-code.

ML(Machine Learning) flipped this around. Instead of writing rules, you show a program enough labelled examples that it figures out the rules itself. Photos labelled "cat" or "dog" → it learns to tell cats from dogs. That same idea, scaled up to billions of examples and billions of internal numbers, gives you what we call AI today. LLMs are one specific kind of ML — applied to language.

So how does it figure things out? A learning algorithm nudges the program's internal numbers, billions of times, until its outputs start to match the labels in the data. Those internal numbers are called weights (or parameters) — there are billions of them in a modern LLM, and together they hold every pattern the model picked up. When training finishes, that giant bag of numbers is the model.

What AI is good at — and what it isn't

The questions everyone has

Try it yourself (no coding, ~5 minutes)

1. Go to claude.ai or chatgpt.com. Sign up free.
2. Ask: "Explain how a microwave heats food, like I'm 8 years old."
3. Then: "Now explain it like I'm a physicist."
4. Then: "Now explain it as a haiku."

Same model, three different outputs — you changed the prompt, the rest adapted. That's the whole interaction. Lesson 1 explains how it does that.

Frequently asked questions

Is this AI lesson beginner friendly?

Yes. It introduces AI in plain language first and does not require coding knowledge.

What will I understand after this lesson?

You will understand the difference between AI, ML, and LLMs, plus how modern AI tools are used in real workflows.

Further reading (no jargon, video-first)

• ChatGPT vs Claude for work (2026) — practical model selection guide.

• 3Blue1Brown — But what is a neural network? — beautifully animated.

• Andrej Karpathy — Intro to Large Language Models — best one-hour overview, plain English.

• 3Blue1Brown — But what is a GPT? — visual intuition for chatbots.

Next: Lesson 1 — What is an LLM, really? Goes one level deeper into the chatbot side.