W01Full seminar3 hours

Introduction to AI Ethics

ANI, AGI, and the moral landscape

What kind of thing is AI—and who gets to say?

We begin by refusing the idea that an AI system is only a model. Students map the stories shaping AI discourse, locate values in technical choices, and use four ethical frameworks on the same cases.

Join the live room

Four ethical lenses arranged around an AI decision — Visual field note · Week 1 · use this diagram to keep the model inside its surrounding system.

Before class

Read one narrative camp charitably. Bring one current AI headline and a two-sentence account of the policy your author would derive from it.

After class

Write the seminar charter in your own words: one norm you expect to find easy, one you expect to find difficult, and how the room should hold you to it.

The promise

By the end of this room…

01Distinguish a technical artifact from the socio-technical system around it.
02Identify how cultural narratives change the policy derived from the same evidence.
03Apply consequentialist, deontological, virtue, and justice-based reasoning.
04Separate fairness, accountability, transparency, privacy, and safety as operational questions.
05Read a canonical AI case against its primary evidence rather than its headline.

Why this week now

Signals, not scene-setting.

The normal-technology, power/empire, existential-risk, and bubble camps now produce distinct engineering and policy programs.

The state is no longer only regulating systems; EO 14319 turns a definition of ideological neutrality into procurement policy.

Stanford's 2026 evidence records rising incident counts while model transparency remains weak.

Run of show

Provoke → frame → work → argue → synthesize.

Open student room ↗

0:00
provocation
One word, then four corners
Students submit one word for AI, then commit to the narrative camp closest to their current view: normal technology, empire, existential risk, or bubble.
Deliverable · A room map we photograph and revisit in Week 13.Live activity · narrative camp
0:15
frame
The model is not the system
Definitions in tension; the socio-technical layer model; the neutrality fallacy; and the course's recurring move from an output to the institution that made it consequential.
0:45
discussion
One headline, four policies
Pairs receive the same current headline, identify its implicit narrative, and state the policy each camp would derive from it.
Deliverable · A 100-word pair post with one evidence link.Live activity · headline policy
1:10
frame
Ethics as engineering questions
Consequences as a loss function, duties as hard constraints, virtue as institutional habit, and Rawlsian justice as a distributional stress test.
1:35
break
Break
Ten minutes. Leave the room's unresolved question visible.
1:45
forensics
Framework speed-dating
Four groups run one assigned framework across COMPAS, Gender Shades, Uber ATG, and the Air Canada chatbot, then apply it to one shared current case.
Deliverable · A framework card: duty, evidence, affected party, action, blind spot.Live activity · framework friction
2:25
controversy
Can neutrality be specified?
Assigned sides debate whether the federal procurement order is bias mitigation, switch sides, then draft a defensible neutrality test—or explain why one cannot exist.
Live activity · neutrality sac
2:50
synthesis
What would move you?
Ratify the seminar norms and name the evidence that could move you out of your opening narrative camp.
Live activity · week 1 exit