Activity 3 - The Tells
Now you know what the machine is doing, you can predict where it will fail. It fails where a picture needs to be right rather than to look right. Five places.
What happens in Activity 3
An 8-minute Phase 3 - Investigation activity, the visual-literacy core of the lesson. Five tell cards (name, where to look, why it happens), a tap-to-pair matching task, then four images shown one at a time with a zoom toggle and tap-to-drop numbered evidence markers, a person/computer judgement, a tell selection, a 1-3 confidence rating, and feedback revealing the designers' evidence after every image.
🎯 Learning objectives
Name five signs an image may have been generated, and where to look for each. Mark evidence on the image itself, not just assert it. Accept the signs are probabilities - a real photograph can look odd, a good generated image can pass.
🖼️ On the images
Images are shown as labelled, described placeholder tiles rather than real photographs, matching the approach used throughout this codebase. The marker-drop, zoom, matching and judgement mechanics are fully real and interactive; only the visual source is simulated. Of the four images, one is a real photograph that looks odd (teaching that tells are probabilities, not proof) and one is a generated image that passes every visible tell (teaching that looking eventually runs out).
- Junior works with three tells (hands, writing, repeats) and four clear-case images
- Zoom is fully keyboard-operable; markers can be placed with keyboard focus and Enter
- Voice narration available for every tell card and question
- Senior gets all five tells and two hard-case images, plus an extra prompt: one of these four cannot be decided by looking - which one, and what would you need instead?
- Print-pack card 3 has the five tells on one side, four large images with an enlarged detail on the other
- Students circle evidence with a pencil instead of dropping markers
- Answers live on a separate teacher sheet