Run an end-to-end ethical review on an AI feature — as you would in a design critique, but with the Module 04 checklist.
Choose a high-impact AI feature: recommendations, scoring, chatbot, content generation, price personalisation. Define context: sector, target users, data used.
Run through the full checklist. For each section (transparency, consent, fairness, accessibility, safety) assign: ✅ OK | ⚠️ Gap | ❌ Critical. Document evidence for every ⚠️ and ❌.
Write 3 worst case scenarios: what happens if the model discriminates, hallucinates, is used out of context? Does the interface make it visible? Can the user report it?
Prioritise findings: must-fix pre-launch, should-fix next sprint, monitor post-launch. For each must-fix, propose a concrete UX solution.
Prepare a one-pager for stakeholders: executive summary (5 lines), top 3 risks, recommendations, "no AI" alternative if applicable.
Ethical design review report with completed checklist, 3 worst cases, priorities and stakeholder one-pager.
Which finding is hardest to get the team to accept? Is there a must-fix you could implement yourself without waiting for approvals? Will you keep this checklist for future projects?