What is Machine Learning? A Designer's Introduction

The AI Family Tree

Before diving into machine learning, it helps to see where it sits in the bigger picture:

AI → Machine Learning → Neural Networks → Deep Learning

Think of it like a set of nested circles. Artificial Intelligence is the broadest concept — any system that mimics human intelligence. Machine Learning is a subset of AI. Neural Networks are a subset of ML. And Deep Learning is a subset of neural networks that uses many layers to learn complex patterns.

So, What is Machine Learning?

Here's the classic definition:

A field of study that gives computers the ability to learn without being explicitly programmed.

In simpler terms — instead of writing exact rules for a computer to follow ("if water pH is below 6.5, mark as polluted"), you give it lots of examples and let it figure out the rules on its own.

A More Formal Way to Think About It

A computer program is said to learn from:

• Experience (E) — the data you feed it
• Task (T) — what you want it to do
• Performance (P) — how you measure success

If performance at the task improves with more experience, the program is learning.

A Concrete Example

Imagine you're building a system to test water quality:

More experience = better performance. The more labelled samples you feed the system, the better it gets at classifying new ones.

Key Terms You'll Hear Everywhere

• ML Algorithm — the method or recipe used to learn from data (e.g., linear regression, decision trees)
• Training Data — the experience you feed the algorithm (your labelled examples)
• Training — the process of the algorithm learning patterns from training data
• ML Model — the output of training; the learned "brain" that can make predictions on new data

The flow looks like this:

ML Algorithm + Training Data → ML Model

Two Types of Training Data

Labelled Data — each data point has a known target value. Example: photos of cats and dogs, each tagged with what they are. The label is the thing you want the model to predict.

Unlabelled Data — you only have features/attributes, no labels. Example: a dataset of customer behaviours with no predefined categories. The model has to find patterns on its own.

This distinction is fundamental and leads directly to the main types of machine learning.

Types of Machine Learning

1. Supervised Learning

The most common type. You provide labelled data, and the model learns a function that maps inputs to outputs:

y = f(x)

• X = features (inputs) — what you know
• Y = labels (outputs) — what you want to predict
• f = the learned function (your model)

Supervised learning breaks down into two sub-types:

A quick rule: if the answer is a category, it's classification. If the answer is a number, it's regression.

2. Unsupervised Learning

No labels. The model explores the data and finds hidden structure — clusters, patterns, anomalies. Think of it as the model organising your messy desk drawer into groups without being told how.

3. Reinforcement Learning

The model learns by trial and error, receiving rewards or penalties. Think of how you might train a pet — good behaviour gets a treat. This is how game-playing AIs and robotics systems often learn.

Why This Matters for Designers

As designers increasingly work with AI-powered products, understanding these fundamentals helps you:

• Ask the right questions during product discovery
• Design better data collection experiences (you're designing the "E" in the learning process)
• Understand model limitations and communicate them in your UI
• Bridge the gap between design and engineering teams working on ML features

More features in your data means more computing power needed — a practical constraint that affects product decisions you'll be part of making.