Lesson plan · 60 minutes
Lesson plan: the risk-based approach
The AI Act's central idea — match the rules to the risk of the use — taught through the exercise that makes it tangible: sorting real-feeling scenarios and defending the sort.
01Audience, duration, objectives
- Target audience — secondary (15+), university, professional training
- Duration — 60 minutes
- Objectives — students can name the four risk levels, sort unfamiliar scenarios into them, and articulate which facts drive a scenario up or down the ladder
- Preparation — prepare 12 scenario slips (from the activities page); review risk categories
02Lesson flow
- Baseline (5') — three mundane examples (spam filter, playlist, spellcheck): establish that most AI is minimal-risk. This inoculates against "everything is banned".
- The ladder (10') — present the four levels with one example each; emphasise that the level attaches to the use in context.
- Sort (20') — pairs sort the 12 slips and flag their two hardest. Circulate; do not correct yet.
- Defend (15') — pairs present their hardest calls; the class challenges with facts ("what if the score only affects queue order?"). Reveal intended levels only at the end.
- Verify in the game (10') — one ambiguous game case (e.g. the adaptive-classroom case) played together on the projector: the class votes each decision step.
03Assessment and reflection
Each student writes the "one detail that changes everything" for a scenario of their choice — the single most reliable indicator they grasped context-dependence. Collect and read three aloud.
04Extension
Assign one focused page per group — prohibited practices, high-risk systems, transparency, GPAI — and have each group teach its level in 3 minutes next lesson.