Unit Overview

Lesson	Title	Key Focus	Time
1	Your First Model	2-class model (circle vs triangle), intro to ML	45–60 min
2	Bias & Fairness	Background trick, biased vs balanced datasets	45–60 min
3	Training vs Testing	TRAIN/TEST split, accuracy calculation	45–60 min
4	Three-Class Model	Multi-class, confusion, model improvement	45–60 min
5	Product Design	Own product idea, prototype, evaluation	45–60 min
6	Showcase & Reflection	Presentation, AI Mythbusters, ethics	45–60 min

Standard Lesson Structure (60 minutes)

Phase	Time	Notes
Starter	5 min	Use starter slides from the deck. Quick discussion or think-pair-share.
Teacher intro / demo	5–10 min	Brief demo on the board. Show the key steps for that lesson.
Main tasks	30–35 min	Students work through the step-by-step instructions. Circulate and support.
Plenary / Exit ticket	5–10 min	Students complete the exit ticket. Quick review of key learning.

Condensed Lesson (45 minutes)

If you only have 45 minutes, reduce the main task time to 20–25 minutes. Consider setting the challenge task as homework. Skip the teacher demo if students are comfortable with Teachable Machine by Lesson 3+.

Supporting Lower-Ability Students

• Pre-made dataset: Provide the demo dataset ZIP so students don't have to draw all images from scratch.
• Paired work: Partner a struggling student with a more confident peer.
• Reduced expectations: Accept 5–8 images per class instead of 10. Focus on the process, not the accuracy score.
• Printed help sheet: Give the help sheet as a laminated handout for students to refer to.
• 1:1 demo: Show the first few steps on their screen before they continue independently.
• Sentence starters: For exit tickets, provide sentence starters on the board (e.g. "Machine learning is when...").

Challenging Higher-Ability Students

• More classes: Ask them to build 4- or 5-class models.
• Tricky test data: Draw ambiguous or deliberately confusing images to test model limits.
• Research: Ask them to research a real-world example of AI bias and write a paragraph about it.
• Accuracy analysis: Have them create a proper confusion matrix (which classes get confused with which).
• Peer teaching: Ask confident students to help others and explain concepts.

Issue	Cause	Fix
Teachable Machine won't load	Browser or network issue	Use Chrome or Edge. Clear browser cache. Check the school firewall isn't blocking the site.
Training freezes	Student switched tabs	Explain they must keep the tab active. If frozen, refresh and re-upload.
All predictions wrong	Images in wrong classes	Check the upload — students may have put circles in the triangle class by mistake.
Very low accuracy	Too few images / too similar	Add more images with more variety. Minimum 8 per class, aim for 10–15.
Students can't save images properly	Using wrong file format or location	Demo how to use File > Save As > PNG. Show them the correct folder path.
Paint canvas too large/small	Default canvas size varies	Resize canvas: File > Properties > set to 400x400 pixels.
Student lost their work on the website	Different computer or cleared localStorage	Data is stored in the browser's localStorage. If they switch computers, data won't transfer. Remind them to use the same machine.

• Exit tickets (every lesson): Check understanding of key vocabulary and concepts. Answers are saved to localStorage — you can view them on individual machines or ask students to read them out.
• Results tables: Look at accuracy scores and whether students can explain why results are good or bad.
• Lesson 5 product design: Assess the design sheet, model accuracy, and evaluation quality. This is the main formative assessment piece.
• Lesson 6 showcase: Assess communication skills, depth of reflection, and understanding of AI ethics.
• Challenge tasks: Use these to identify students working above expected level.

This unit supports the following areas of the KS3 Computing curriculum:

• Understand several key algorithms: Students explore how classification algorithms learn from data.
• Understand simple Boolean logic: Prediction is essentially a classification decision (is it A or B?).
• Use technology safely, respectfully, and responsibly: Covered through the bias, fairness, and ethics activities.
• Be able to evaluate the effectiveness of computational solutions: Students evaluate their models' accuracy and suggest improvements.