Activity 2 - What Is the Machine Actually Doing?
One accurate mental model. Every one of the five tells in Activity 3 is a consequence of this - so it has to come first, and it has to be right.
What happens in Activity 2
A 7-minute Phase 2 - Encounter activity. Four explanation screens (training data, a prompt demo, noise turning into an image in five steps, and the hinge sentence held on screen), a tap-to-order task with five cards and three attempts before a guided reveal, then two ungraded check questions.
🎯 Learning objectives
Describe in your own words what an image generator does. Explain what training data is and where it came from. Connect the way the machine works to the mistakes it makes.
⚠️ Accuracy, on purpose
This activity deliberately never says the generator "copies" or "collages" existing images, and never says it "thinks", "imagines" or "decides" - it measures, matches, produces and builds. It also never claims the model is "always wrong about hands" - the tells in Activity 3 are probabilities, not rules.
- Junior sees three explanation screens (the noise-to-image steps are watched rather than tapped through) and a simplified four-card ordering task
- Voice input on the C1 field; extended time from the accessibility profile
- Senior gets all four screens with the scrubbable noise animation, five ordering cards, and an extra prompt: the same prompt produces a different image every time - why?
- Print-pack card 2 is a one-page comic-strip explainer with the same four steps
- The ordering task becomes five cut-out cards to arrange on the desk
- The prompt demo can be run once from the projector for the whole class