Forecasts should be scored against observed reality; vividness is not calibration.
Capstone & Synthesis
Final defense and the future of AI
Can you defend an ethical position under fire about systems that did not exist when the course began?
Students defend complete ethical audits, red-team one another's evidence, revisit their Week 1 narrative positions, and leave with a durable analytical practice rather than a frozen list of rules.
Join the live roomBefore class
Submit the final audit, ethical matrix, pre-mortem, regulatory map, and one page on which claims could be wrong within twelve months.
After class
Keep the one-word Week 1 card. Write the missing Week 14 you think the next edition will need.
The promise
By the end of this room…
- 01Synthesize the course into a coherent analytical method.
- 02Defend claims with technical precision, normative clarity, and calibrated uncertainty.
- 03Red-team an audit across evidence, stakeholders, law, and remedy.
- 04Evaluate future scenarios against their assumptions and track record.
Why this week now
Signals, not scene-setting.
Frontier-safety, present-harm, model-welfare, and accelerationist agendas now compete through institutions and budgets.
The most durable professional skill is knowing which facts could change and how the recommendation survives when they do.
Run of show
Provoke → frame → work → argue → synthesize.
- 0:00provocation
Commit before the concepts
Which Week 1 narrative camp is closest to your view now?
Live activity · week 13 opening - 0:15frame
The audit defense standard
Forecasts should be scored against observed reality; vividness is not calibration.
- 0:45discussion
Reading tension
Student leaders present the assigned readings as a clash of defensible positions, then moderate questions that expose the hidden assumptions.
- 1:10frame
Forecasts as accountable arguments
Frontier-safety, present-harm, model-welfare, and accelerationist agendas now compete through institutions and budgets.
- 1:35break
Break
Ten minutes. Leave the room's unresolved question visible.
- 1:45forensics
Capstone red-team defenses
Role-based groups work the anchor case through technical, legal, stakeholder, and normative lenses.
Deliverable · A two-minute finding with evidence, uncertainty, and an actionable remedy.Live activity · week 13 forensics - 2:25controversy
Futures fishbowl: where is the field at mid-career?
Assigned positions, side-switch, and a joint recommendation that names the value or stakeholder it leaves exposed.
Live activity · week 13 controversy - 2:50synthesis
Re-vote and leave a trace
Repeat the opening vote, inspect what moved, and submit the strongest argument you still reject.
Live activity · week 13 exit
Case room
Evidence before opinion.
Student ethical audits
Which recommendation survives adversarial evidence and changed law?
AI 2027
Which assumptions are measurable, and what would count as being wrong?
AI welfare
How should moral uncertainty allocate scarce attention?
Reading stack
Read the tension, not the bibliography.
- 01CoreClosing the AI Accountability Gap ↗
Raji et al.
- 02CurrentInternational AI Safety Report 2026 ↗
International expert panel
- 03CurrentAI 2027 ↗
AI Futures Project
- 04OptionalTaking AI Welfare Seriously ↗
Long, Sebo, Chalmers et al.
Evidence ledger
Every case has a receipt.
5 primary, scholarly, or first-party sources
Closing the AI Accountability Gap
Raji et al.'s end-to-end internal algorithmic auditing framework and the spine of the capstone audit.
Audit framework ↗Artificial Intelligence Risk Management Framework 1.0
The primary source for Govern–Map–Measure–Manage and lifecycle risk ownership.
Risk-management framework ↗International AI Safety Report 2026
A shared technical evidence base for frontier capabilities, risks, safeguards, and uncertainty.
International evidence synthesis ↗AI 2027
A prominent forecast to analyze as a scored scenario rather than as a prophecy.
Scenario ↗Taking AI Welfare Seriously
A deliberately difficult final-week reading about uncertainty, moral status, and precaution.
Research agenda ↗