Embedded Ethics Module



Title: Ethics of Algorithms for Resource Allocation

Duration: 2 hours

Time & date: 3-5pm on Nov 25 (LEC 0201) and 1-3pm on Nov 30 (LEC 0101)

Mode: Zoom call for the lectures (please note that the module will be Zoom-only; there will be no in-person component)

Collaboration: Course instructor (Nisarg Shah), Ethics TA (Deepanshu Kush), Embedded Ethics Team (Steven Coyne, Diane Horton, David Liu, Emma McClure, Sheila McIlraith, Nina Wang)


Before the module:
  • [0.5% Grade] Complete the pre-module survey by Nov 1, 11:59pm ET.
  • [Not graded, but highly encouraged] Watch two introductory videos about fairness in resource allocation: one from a philosophical viewpoint (prepared by Steven Coyne from the Embedded Ethics Team) and the other from an algorithmic viewpoint (prepared by Deepanshu Kush, slides available here).
  • [2% Grade] Submit your solution to the pre-module homework to MarkUs by Nov 25, 2pm ET.

Module overview:

In this module, we will consider ethics of algorithm design, illustrated in the domain of resource allocation. In particular, we will focus on a vaccine distribution problem, set up in the pre-module homework. We will begin with a natural modeling of this problem via network flow such that maximizing the amount of flow maximizes the number of vaccine doses delivered.

In a group exercise (breakout activity), we will first brainstorm various ethical issues with this natural modeling. Then, in another group exercise (breakout activity), we will start to design a more ethical solution from scratch by first identifying the various stakeholders and their role in this problem. This will be followed by a number of Zoom polls asking to weigh the ethical issues associated with alternative solutions.

Before concluding, we will review conceptual and technical challenges involved in designing algorithms for resource allocation. Ethical issues not emphasized by or investigated in this module will be noted.


Module goals:
  1. Introduce students to the topic of algorithmic fairness.
  2. Demonstrate connections between concepts learned in class (e.g., network flow) and real-world problems (e.g., vaccine distribution).
  3. Explain philosophical and algorithmic approaches to fairness and distributive justice.
  4. Equip students with tools to reason about fairness in algorithmic decision-making.
  5. Practice thinking and communicating about what fairness requires in a given real-world context.
  6. Practice identifying stakeholders and thinking from their perspectives.

Module agenda:
  • [02 min] Introduce EE team, summarize goals of the module
  • [02 min] Introduction to Breakout Activity 1
  • [20 min] Breakout activity 1: Technical Specification of Problem
  • [15 min] Discuss results
  • [10 min] Break
  • [04 min] Introduction to Breakout Activity 2
  • [20 min] Breakout Activity 2: Stakeholders and Ethics
  • [15 min] Discuss results
  • [15 min] Polls about policies (three polls + discussion)
  • [05 min] Takeaways: technical challenges, conceptual challenges, what we didn't cover
  • [02 min] Conclusion & feedback

After the module:
  • [0.5% Grade] Complete the post-module survey (link TBA) by (due date TBA).
  • [2% Grade] Submit your solution to the post-module homework to MarkUs by (due date TBA).