All courses

Beginner · module B2

How Artificial Intelligence Works: Data and Learning

By the end: Describe how an AI system learns from data, train and test a simple model in the browser, and explain why data quality matters

Covers
Data, how computers store it, and how machines learn from examples, without any programming
Contents
5 chapters, 23 topics
Time
About 3 hours
Level
Beginner
Your progress
0 / 23

This module builds on Introduction to Artificial Intelligence. We suggest looking at it first, but the choice is yours.

Chapter 1 Data: The Foundation of AI

  1. 1.1

    What Is Data?

    Facts collected to answer a question, and the difference between data and information

    Unplugged: collect data from a group to plan an event

    8 min Read Activity
  2. 1.2

    Types of Data

    Numbers, text, images, audio and video; structured and unstructured data

    Interactive: sort data cards into types

    8 min Read Activity
  3. 1.3

    Collecting and Organising Data

    Surveys, sensors and observation; tables, tally marks, pictographs and bar graphs

    Interactive: turn survey answers into a table and a bar graph

    10 min Read Activity
  4. 1.4

    How Computers Store Information

    Everything becomes numbers; bits, bytes and binary in simple terms

    Unplugged: binary cards

    10 min Read Activity
  5. 1.5

    How Computers See Images

    Pixels, resolution and colours stored as numbers

    Interactive: build a picture pixel by pixel

    10 min Read Activity
  6. 1.6

    How Computers Hear Sound

    Sound waves turned into numbers that a computer can store

    Interactive: see a voice as a wave

    6 min Read Activity

Chapter 2 Learning from Examples

  1. 2.1

    Rule-Based Systems and Learning Systems

    Two ways to make a machine smart: writing rules or learning from examples

    Interactive: write rules for a spam filter, then watch them fail

    10 min Read Activity
  2. 2.2

    Learning from Examples

    How a child learns to recognise a dog, and how a machine learns the same way

    Unplugged: a group trains a "human model"

    8 min Read Activity
  3. 2.3

    Examples, Labels and Features

    Labels are the right answers; features are the clues a model uses

    Interactive: choose the clues that separate apples from oranges

    10 min Read Activity
  4. 2.4

    Training a Model

    Showing a model many labelled examples so it can find a pattern

    Interactive lab: train an image model in the browser

    12 min Read Activity
  5. 2.5

    Testing a Model

    Checking a model on new examples it has never seen

    Interactive lab: test your model with new pictures

    8 min Read Activity
  6. 2.6

    Predictions and Confidence

    A model's answer, and how sure it is about it

    Interactive: read the confidence bars

    8 min Read Activity
  7. 2.7

    Learning from Mistakes

    Feedback, more examples and better examples make a model improve

    Interactive: fix a model that confuses two classes

    8 min Read Activity

Chapter 3 Three Ways Machines Learn

  1. 3.1

    Supervised Learning

    Learning from examples that come with correct answers

    Interactive: sort messages using labelled examples

    8 min Read Activity
  2. 3.2

    Unsupervised Learning

    Finding groups and patterns when there are no answers

    Interactive: group mystery objects by similarity

    8 min Read Activity
  3. 3.3

    Reinforcement Learning

    Learning by trial and error through rewards and penalties

    Interactive game: guide a robot through a maze with rewards

    8 min Read Activity

Chapter 4 Quality of Data and Models

  1. 4.1

    Measuring Accuracy

    Counting correct answers to see how good a model is

    Interactive: score a model on ten test pictures

    8 min Read Activity
  2. 4.2

    Learning Compared with Memorising

    Why a model that only memorises fails on new examples

    Discussion: practice questions compared with a real exam

    6 min Read Activity
  3. 4.3

    Good Data and Bad Data

    Too little data, wrong labels, missing values and one-sided data

    Interactive: find the problems in a dataset

    8 min Read Activity
  4. 4.4

    How Data Shapes What AI Learns

    Data that leaves people out produces a model that is unfair to them

    Unplugged: the biased bag of examples

    6 min Read Activity

Chapter 5 The Complete Learning Cycle

  1. 5.1

    The AI Learning Cycle: Data, Model and Feedback

    The full loop from data to prediction to improvement

    Interactive: run the learning cycle step by step

    8 min Read Activity
  2. 5.2

    Project: Build and Test Your First AI Model

    Plan, train, test and improve a picture classifier, with no programming

    Guided project in the browser with a test record

    20 min Read Activity
  3. 5.3

    Module Review and Assessment

    Recap of the module with a scored quiz

    Module quiz with explanations

    10 min Read Activity