Beginner · module B8
Computational Thinking and the AI Project Cycle
By the end: Break a problem into steps, express a solution as a flowchart or decision tree, and plan an AI project from problem to deployment
- Covers
- How to think through a problem step by step, and the six stages used to plan any AI project
- Contents
- 4 chapters, 21 topics
- Time
- About 3 hours
- Level
- Beginner
- Your progress
- 0 / 21
This module builds on How Artificial Intelligence Works: Data and Learning. We suggest looking at it first, but the choice is yours.
Chapter 1 Computational Thinking
- 1.16 min Read Activity
Decomposition
Breaking a large problem into smaller parts
Unplugged: plan a school event step by step
- 1.26 min Read Activity
- 1.36 min Read Activity
- 1.48 min Read Activity
Algorithms
Precise steps that solve a problem, such as a recipe
Unplugged: write instructions a partner must follow exactly
- 1.56 min Read Activity
Chapter 2 Algorithms and Logic
- 2.18 min Read Activity
- 2.28 min Read Activity
- 2.310 min Read Activity
Decision Trees
Making decisions with a series of if-then questions
Interactive: build an animal-guessing tree
- 2.48 min Read Activity
Optimisation: Finding the Best Solution
Shortest paths and best choices
Unplugged: string map for the shortest route
Chapter 3 The AI Project Cycle
- 3.18 min Read Activity
Overview of the AI Project Cycle
Problem scoping, data acquisition, data exploration, modelling, evaluation and deployment
Interactive: order the six stages
- 3.210 min Read Activity
Problem Scoping and the 4Ws Canvas
Who, what, where and why of a problem, and the problem statement
Template: fill the 4Ws canvas
- 3.38 min Read Activity
Stakeholders and Design Thinking
Understanding the people affected, with empathy maps
Activity: draw an empathy map
- 3.410 min Read Activity
Data Acquisition and System Maps
Choosing data features and sources, and drawing how parts connect
Activity: draw a system map
- 3.510 min Read Activity
Data Exploration
Using charts to understand data before modelling
Spreadsheet lab: explore a small dataset
- 3.68 min Read Activity
Modelling: Rule-Based and Learning-Based
Choosing between written rules and a learned model
Card sort
- 3.710 min Read Activity
Evaluation: Measuring Mistakes
True and false positives and negatives, and why not all mistakes are equal
Interactive: evaluate a spam filter
- 3.86 min Read Activity
Deployment
Putting a model to use, and checking it keeps working
Case study: preventing avoidable blindness
- 3.910 min Read Activity
When a Model Stops Working Well
Data drift, fresh test examples and planned human checks
Scenario: the waste sorter in the monsoon
Chapter 4 No-Code AI
- 4.115 min Read Activity
Building Models with No-Code Tools
Training image, sound and pose models without programming
Interactive lab: build a sound classifier
- 4.212 min Read Activity
Testing an AI Tool: Can We Trust It?
Checking a chatbot and a prediction tool for accuracy and fairness
Lab: run a test plan and record results
- 4.310 min Read Activity