Artificial Intelligence is just a computer program that can do tasks we normally think require human intelligence — like recognising faces, understanding speech, or playing chess.
Rule-based programs
A programmer writes every single rule. "If the email contains 'free money', mark it as spam." Works for simple stuff. Breaks down for anything complex.
AI / Learning systems
Instead of writing rules, you show the computer thousands of examples and let it figure out the rules itself. This is what machine learning is.
Teaching a child what a "dog" is — you don't give them a textbook definition. You point at many dogs and say "dog!". Eventually they just… know. AI learns the same way, from examples.
Machine Learning = giving a computer data, and letting it learn patterns from that data — without being explicitly programmed with rules.
Collect Data
Thousands (or millions) of examples. For spam detection: emails labelled "spam" or "not spam".
Train a Model
The algorithm studies the examples and builds an internal map of what patterns lead to which outcomes.
Make Predictions
Show the model something new it's never seen. It uses what it learned to make a prediction.
Imagine a doctor who's seen 10,000 X-rays. They can look at a new one and spot a problem — not because they memorised a rulebook, but because they've seen enough examples to feel what's normal and what isn't. An ML model works the same way.
Think of learning to shoot a basketball. You throw, miss, adjust your angle, throw again. Each attempt makes you slightly better. A machine learning model does exactly this — millions of times per second.
You're the trainer!
Below are animals. Click each one to label it. Mark the 🐶 dogs as ✅ and everything else as ❌. After labelling, hit "Train" to see the AI learn!
When an AI model trains, it builds a list of weights — numbers that say how important each feature is for making a prediction.
How the AI decides "spam or not spam"
These bars show how strongly each word influences the spam verdict. Click "Run Training" to see the weights update as the AI learns!
The model doesn't know what "free money" means. It just learns: "every time these pixels/words appear, the answer tends to be X." Intelligence from statistics, not understanding.
A neural network is a stack of layers. Each layer takes in numbers, does a calculation, and passes new numbers to the next layer. The final layer outputs the answer. Deep learning just means using many layers stacked together.
Each circle is a "neuron". Glowing connections are active (high weight). Data flows left → right.
Input Layer
Raw data goes in. Pixels of an image, words in a sentence, temperature readings — whatever you're predicting from.
Hidden Layers
These detect patterns at increasing levels of complexity. First layer: edges. Next: shapes. Next: faces. Then: identities.
Output Layer
The answer. "Cat (92% confident)", or a word prediction, or a price estimate — whatever the task demands.
Machine learning is already everywhere. Here are real examples, and what's happening under the hood.
You've now seen the whole picture: Data goes in → an Algorithm finds patterns → a Model is built → it makes Predictions. Neural networks are just a powerful version of this idea, with many layers stacked together.
You've got the fundamentals!
You now understand what AI is, how ML trains from data, what neural networks are, and where it all shows up. That's the foundation that even most engineers start from.