Unit 6 Project: Ethical AI in Practice
Ethical AI in Practice
Apply everything from Unit 6 to a real-world governance challenge. You'll analyze scenarios, assign roles, design mitigations, and draft a governance plan — demonstrating mastery across all five lessons.
Your mission: be the AI governance expert
You've completed all five lessons of Unit 6. Now it's time to put those skills to work. This capstone presents four real-world-style governance scenarios. For each one, you'll need to draw on everything you've learned — ethical risk identification, responsible AI principles, global regulations, privacy compliance, and stakeholder roles.
Work through each case study tab below. Complete the response tasks in each section, then submit for scoring and feedback. There are no trick questions — this is about demonstrating you can think like an AI governance professional.
Lesson 6.1 (Ethical Risks) · Lesson 6.2 (Responsible AI Principles) · Lesson 6.3 (Global AI Regulations) · Lesson 6.4 (Privacy & Data Compliance) · Lesson 6.5 (Governance, Mitigation & Stakeholder Roles)
- Case Study A: Healthcare AI Bias Crisis
- Case Study B: EU Expansion & Privacy Compliance
- Case Study C: Hiring AI Under Regulatory Scrutiny
- Case Study D: Governance Structure Design
Four case studies · 25 points each
Case Study A: The HealthPredict Diagnostic AI Crisis
Unit 6.1 (Ethical Risks) · Unit 6.2 (Responsible AI Principles) · Unit 6.5 (Governance Roles)
Hint: Think bias, transparency, accountability — and connect to specific principles like fairness, safety, inclusiveness.
Roles to consider: Model Owner, Data Scientist / ML Engineer, Compliance Officer, DPO, Ethics Committee, CISO, AI Product Manager, Clinical Leadership.
Consider: immediate risk containment, bias mitigation strategies, data remediation, retraining, monitoring changes, and governance structure improvements.
Case Study B: RetailMax Expands to the EU — Privacy Reckoning
Unit 6.3 (Global AI Regulations) · Unit 6.4 (Privacy & Data Compliance) · Unit 6.5 (Governance)
Think: lawful basis, transparency, data minimization, automated decision-making rights, DPO requirement, DPIA, consent, data processing agreements.
This is a communication exercise — be clear, factual, and persuasive without being condescending.
Prioritize by legal risk and launch dependency. Include: DPO appointment, DPIA, lawful basis documentation, privacy notices, data processing agreements, user rights infrastructure.
Case Study C: HireLogic — Regulatory Scrutiny and a Discrimination Complaint
Unit 6.1 (Ethical Risks) · Unit 6.3 (Global Regulations) · Unit 6.4 (Privacy) · Unit 6.5 (Governance)
Consider: EEOC / Age Discrimination in Employment Act (ADEA), NYC Local Law 144, GDPR (if any EU clients), Title VII, Illinois AIVIA, and any relevant AI Act provisions.
This is a live crisis. Balance legal compliance, client obligations, potential harm to job seekers, and business continuity. What can't wait vs. what can be sequenced?
Address: bias testing process, model cards, ethics committee structure, client contracts, regulatory monitoring, ongoing audit schedule. Be specific about what each element would look like.
Case Study D: Building AI Governance from Scratch
Unit 6.2 (Responsible AI Principles) · Unit 6.5 (Governance Lifecycle & Roles) — Synthesis Case
Consider: EU AI Act high-risk categories, ECOA/fair lending law, GDPR automated decision rights, CFPB AI guidance, and the consequences of errors for affected individuals.
Balance comprehensiveness with practicality — this is an 800-person company, not a tech giant. What's the minimum viable governance structure that still provides real protection?
Reference the lifecycle stages: problem definition → data collection → model development → testing & validation → deployment → monitoring & update. Include who is responsible at each stage.
How each case study is scored (25 points each)
| Criterion | Exemplary (Full marks) | Satisfactory (Partial) | Points |
|---|---|---|---|
| Risk / Issue Identification | Accurately identifies all major risks, violations, or gaps with specific detail. Names correct laws, principles, or articles. | Identifies major issues but misses nuances or lacks specificity on applicable laws/principles. | 8 pts |
| Stakeholder / Role Assignment | Correctly assigns each action to the right role with specific, actionable responsibilities. No role confusion. | Mostly correct assignments; may have one role confusion or some actions assigned to wrong stakeholders. | 8 pts |
| Mitigation / Plan Quality | Comprehensive, practical plan with clear sequencing, specific actions, and prevention mechanisms. Shows systems thinking. | Plan covers main points but may be generic, miss some elements, or lack sequencing and prevention focus. | 7 pts |
| Vocabulary & Concept Application | Consistently uses accurate Unit 6 terminology in context (bias audit, DPIA, model drift, disparate impact, XAI, etc.). | Uses some terms correctly; may be inconsistent or occasionally use terms imprecisely. | Embedded above |
After submitting each case, read the model answer guidance carefully. Governance rarely has a single correct answer — the quality of your reasoning and specificity of your recommendations matters as much as covering the right topics. Would a DPO or compliance officer find your plan actionable?