Module 2: Bias, fairness, and representational harm

Module 2: Bias, fairness, and representational harm#

AINS6005 — AI Ethics, Law & Policy

Essential Question#

How do datasets and objectives encode inequity?

Scenario#

an AI governance board reviewing a proposed high-impact AI deployment

Stakeholders: governance chair, legal counsel, affected-user advocate, product owner, and compliance officer

Core Moves#

  • Define the decision boundary

  • Compare baseline and alternative

  • Interpret evidence and assumptions

  • Identify failure modes

  • Recommend next action

Lab & Assignment#

Audit a model scenario for bias pathways and mitigations.

Artifact: responsible AI review memo with risk register, policy analysis, and redress plan focused on bias, fairness, and representational harm: Audit a model scenario for bias pathways and mitigations.