Teach the AIGP as an eight-session cohort
Each 90-minute session pairs lessons from the course with a hands-on activity and discussion prompts. Learners read the lessons before class and use Practice and Mock exam between sessions.
How to run the cohort
Before each session
Assign the listed lessons. Learners answer each lesson's checks and mark it complete.
In the session
Short recap, then most of the time on the activity. Close with five practice questions answered live.
Tracking
Ask learners to share a screenshot of their Progress page each week; aim for 80% in every domain by session 8.
What AI is and why it needs governing
Pre-reading
Objectives
- Explain the OECD definition
- Map a real system to lifecycle stages
Agenda
- 10 min · Introductions and exam overview
- 25 min · Mini-lecture: AI types and the lifecycle
- 30 min · Activity
- 15 min · Discussion and quiz
Homework
Lessons 1.1–1.2 and their checks.
Activity
Teams pick an AI system they use at work and place each part of its life on the seven-stage lifecycle, noting who owns each stage.
Discussion prompts
- Where in your organization is AI already in use without being called AI?
- Which lifecycle stage has the weakest ownership today?
Risks, principles and the governance program
Pre-reading
Objectives
- Classify harms by affected party
- Design committee roles and an inventory
Agenda
- 20 min · Harms by party
- 20 min · Principles and NIST characteristics
- 35 min · Activity
- 15 min · Debrief
Homework
Lessons 1.3–1.5; Domain I practice set.
Activity
Draft a one-page AI governance charter: committee members, decision rights, risk tiers and the fields of an AI inventory.
Discussion prompts
- Which existing process (privacy, security, model risk) would you extend first?
- How would you stop the program becoming a bottleneck?
Existing law reaches AI
Pre-reading
Objectives
- Apply GDPR Arts. 22 and 35
- Explain disparate impact and FTC enforcement
Agenda
- 30 min · GDPR applied to training and inference
- 25 min · US consumer and anti-discrimination law
- 25 min · Case study
- 10 min · Quiz
Homework
Lessons 2.1–2.2; flashcards.
Activity
Case: a lender adopts an ML underwriting model. Groups list every legal duty triggered (GDPR, ECOA, FCRA, FTC) and the evidence they would keep.
Discussion prompts
- Is a human who approves 99% of AI decisions 'meaningful' review?
- Who is liable when a vendor tool discriminates?
The EU AI Act
Pre-reading
Objectives
- Classify systems into risk tiers
- Assign provider vs deployer duties
Agenda
- 30 min · Scope, roles, tiers
- 25 min · Obligations, GPAI, dates, penalties
- 25 min · Classification drill
- 10 min · Quiz
Homework
Lessons 2.3–2.4; read Art. 5 and Annex III.
Activity
Classification drill: give teams 12 short system descriptions; they assign a tier, the role of each party, and one obligation that applies.
Discussion prompts
- When does a deployer become a provider?
- Which obligations would be hardest for your organization?
Frameworks, standards and the global map
Pre-reading
Objectives
- Use NIST AI RMF functions
- Relate ISO/IEC 42001 to legal duties
Agenda
- 30 min · NIST AI RMF and ISO
- 20 min · US states and global laws
- 30 min · Mapping exercise
- 10 min · Quiz
Homework
Lessons 2.5–2.6; Domain II practice to 80%.
Activity
Map five EU AI Act provider obligations to NIST functions and ISO/IEC 42001 clauses to show how one program can meet several regimes.
Discussion prompts
- Should a company seek ISO/IEC 42001 certification? Who benefits?
- How do you track fast-changing state laws?
Governing development
Pre-reading
- Plan and design: scoping the system
- Data governance and bias
- Building and training with privacy in mind
- Testing, explainability and security
Objectives
- Set lifecycle gates
- Choose fairness metrics and tests
Agenda
- 20 min · Design and data
- 20 min · Bias and fairness metrics
- 35 min · Activity
- 15 min · Debrief
Homework
Lessons 3.1–3.4.
Activity
Teams write the test plan and release-gate checklist for a résumé-screening model, including which fairness metric they choose and why.
Discussion prompts
- Which fairness metric fits hiring, and which fits medical triage?
- Who should be allowed to accept residual risk?
Release, monitoring, generative AI and deployment
Pre-reading
- Release, monitoring and change control
- Governing generative AI
- Assessing the use case and human oversight
- Procurement, third-party risk and contracts
Objectives
- Plan monitoring and change control
- Run vendor due diligence
Agenda
- 20 min · Monitoring and GenAI controls
- 20 min · Oversight and procurement
- 35 min · Role-play
- 15 min · Debrief
Homework
Lessons 3.5–3.6, 4.1–4.2.
Activity
Procurement role-play: one side is a GenAI vendor, the other a deployer. The deployer must get answers to the eight due-diligence questions and agree five contract terms.
Discussion prompts
- What would make you walk away from a vendor?
- How do you make human oversight real at high volume?
Operations, incidents and exam readiness
Pre-reading
Objectives
- Run an AI incident tabletop
- Apply exam strategy
Agenda
- 30 min · Incident tabletop
- 20 min · Exam strategy
- 40 min · Timed mock exam (in class or at home)
Homework
Lesson 4.3; two full mock exams; review misses.
Activity
Tabletop: a deployed chatbot gives customers false medical advice. Teams walk through define, triage, contain, notify, investigate, fix and learn.
Discussion prompts
- Which notification duties apply, and on what clock?
- What would you change in the program afterward?