Introduction to AI — AI Course

What you'll learn

Artificial intelligence is no longer science fiction — it writes, draws, drives, and diagnoses. This course explains how it actually works, from the ground up, with an animation for almost every idea. No PhD required: if you know a little Python and can read a graph, you can follow all of it.

By the end of this first module you'll be able to:

Say what artificial intelligence really means
Tell apart AI, machine learning, and deep learning
Explain why AI suddenly got so good in the last few years
See the whole map of what you're about to learn

What is artificial intelligence?

Artificial intelligence is the effort to make computers do things that normally require human intelligence — recognizing a face, understanding a sentence, planning a route, playing a game. For decades we attempted this by writing the rules ourselves. That works for chess-like problems with clear rules, but it falls apart for messy ones: nobody can write down every rule for "what a cat looks like."

The breakthrough was to stop writing the rules and instead let the machine learn them from examples. That is machine learning, and it is the engine under almost everything people now call "AI."

spam_filter.py

# Two ways to build a spam filter.

# 1) Traditional programming: a human writes the rules.
def is_spam_rules(email):
    if "free money" in email.lower():
        return True
    if email.count("!") > 5:
        return True
    return False           # ...and a hundred more brittle rules

# 2) Machine learning: show a model examples and it LEARNS the rule.
model.fit(emails, labels)        # labels: spam / not-spam
is_spam_ml = model.predict(new_email)

# The ML version was never told what spam looks like —
# it discovered the pattern from data.

Same goal, two philosophies: hand-written rules vs. a pattern learned from data.

Key idea

The core shift of modern AI: we no longer program the solution — we program a system that learns the solution from data. Everything else in this course is a variation on that idea.

AI, machine learning & deep learning

These three words get used interchangeably, but they nest inside one another like Russian dolls. Machine learning is one approach to AI; deep learning is one (extremely powerful) approach to machine learning; and today's generative AI is built almost entirely on deep learning.

The nested family of AI

Artificial Intelligence

Any technique that makes machines act smart — including hand-written rules.

Machine Learning

Systems that learn the rules from data instead of being told them.

Deep Learning

ML with many-layered neural networks that learn their own features.

Generative AI / LLMs

Deep networks that create text, images & code — today's frontier.

Each layer is a subset of the one around it. This course climbs from the outside in.

Why did AI get so good so fast?

Neural networks are decades old — so why the sudden explosion? Three ingredients finally arrived at once. Take any one away and the modern AI boom doesn't happen.

The three ingredients

Data

The internet, phones & sensors produce oceans of text, images, and clicks to learn from.

Compute

GPUs (and TPUs) do the billions of parallel multiplications training needs.

Algorithms

Backpropagation and the Transformer turned raw data & compute into capability.

Note

This is why the same core idea (a neural network) that was a curiosity in the 1990s became world-changing in the 2020s: the idea barely changed, but the data and compute around it grew a million-fold.

The map of this course

We build up in eight phases, each one standing on the last. You can feel the arc: from "what is learning?" all the way to "how does ChatGPT work?"

Eight phases, zero to advanced

Foundations

What AI is, how machines learn, data & the math.

Classical ML

Regression, classification, trees, clustering + gradient descent.

Making models work

Evaluating honestly and beating overfitting.

Neural networks

Neurons, forward propagation & backpropagation.

Deep learning

Depth, CNNs for vision, RNNs for sequences.

Language & Transformers

Tokens, embeddings, attention, the Transformer.

Generative AI & LLMs

How ChatGPT & image generators work.

AI in the real world

Reinforcement learning, ethics & shipping AI.

Every phase turns its ideas into real, animated examples you can play with.

How to read this course

Two things make this different from a textbook:

Play the animations. Almost every idea has an interactive player — press Play, or step through one frame at a time with the arrows. Each step has a caption, so you never have to guess what happened.
Read the code. Every example is in Python — first written from scratch so you see the mechanism, then with real libraries (scikit-learn, PyTorch) so you see how it's really done.

Tip

Don't just watch the animations — predict the next step before you press it. Actively guessing is how these ideas move from "makes sense" to "I could rebuild it."

Recap & quick check

Key takeaways

AI is making machines do things that normally require human intelligence.
The modern approach is machine learning: learn the rules from data instead of hand-coding them.
AI ⊃ machine learning ⊃ deep learning ⊃ today's generative AI — nested subsets.
AI took off because data, compute (GPUs), and algorithms (backprop, Transformers) arrived together.
This course climbs eight phases from 'what is learning?' to how LLMs work — with an animation for nearly every idea.

Quick check

1. What is the key idea that separates machine learning from traditional programming?

2. How do AI, machine learning, and deep learning relate?

3. Which trio of ingredients caused the recent explosion in AI ability?

4. Why did rule-writing fail for tasks like recognizing a cat in a photo?

So the whole game is learning from data. In the next module we open up that black box and see the single loop that every learning system runs. Next up: Module 2 — How Machines Learn.