Module 3 · Level I · Foundations

How machines learn

Data, training and prediction; supervised, unsupervised and reinforcement learning, explained with everyday examples and no maths.

6 lessons · about 1 hour

The module in 48 secondsCaptions on · sound optional

What’s inside

  1. 01
    Learning from examples: data, features and labels

    Machine learning swaps hand-written rules for examples, and three words (data, features and labels) explain how. · 9 min

  2. 02
    Supervised learning: learning with an answer key

    The most common kind of machine learning learns from examples that come with the right answers attached. · 10 min

  3. 03
    Unsupervised learning: finding groups nobody named

    When there are no answers to learn from, machines can still find structure, such as customer groups and odd-one-out events. · 9 min

  4. 04
    Reinforcement learning: trial, error and reward

    Some AI learns by trying things and being rewarded, which is how it beat world champions at games and how chatbots learn manners. · 10 min

  5. 05
    Neural networks: layers of adjustable dials

    The idea behind almost all modern AI, explained as layers of dials that get nudged until the answers come out right. · 10 min

  6. 06
    When learning goes wrong: overfitting and bias

    Machines learn exactly what the data teaches, which is why memorising, bad data and old unfairness all end up in the results. · 11 min

Where this sits on your path depends on your assessment. Some people skip this module outright; others start here. The free placement assessment decides, and the first lesson of every module is free to read. Find your level →

Also at Level I

1What AI actually is
AI, machine learning, deep learning and generative AI, untangled: what each term means, what it doesn't, and why the hype and the fear both miss.2A short history of AI
From Turing and the 1956 Dartmouth workshop through two AI winters, deep learning and the transformer to ChatGPT and agents: how we got here, and what it predicts.4Types and styles of AI
Narrow and general, predictive and generative, rules and learning: the different styles of AI and where you already meet each one.5Meet the models
Language, image, speech and video models; open and closed; large and small; reasoning models: the kinds of model behind the tools, and who makes them.