This is a writing assignment concerning algorithmic bias and fairness. Answer
each of the following ten questions separately. Keep all your answers in two
pages.
1. Algorithms and bias
First, read the following interview with Cynthia Dwork, Gordon McKay Professor of Computer Science
at Harvard University.
Based on what you learned from this interview, answer each of the following
four questions separately in complete English sentences. Be as
specific as possible.
How do algorithms discriminate?
How do historical biases contribute to algorithmic biases?
In her interview, Dwork is asked to talk about the seminal article titled
Fairness through awareness
[1]. In summary, what is the article all about?
Provide an example of fairness through awareness discussed in the
interview.
2. Algorithmic fairness
Next, read the following article by Manish Raghavan, Drew Houston (2005) Career Development Assistant Professor at Masachusetts Institute of Technology.
Based on what you learned from this article, answer each of the following six
questions separately in complete English sentences. Be as specific as
possible.
Raghavan begins by discussing Dwork's pioneering work and states that the
primary focus of this article is group fairness. What is the notion
of group fairness?
What is algorithmic fairness by a diagnostic baseline?
As discussed on pp. 90–92, the impossibility theorems for fair
decision making, orignally published in [2], are
the results of mathematically generalizing a specific example in criminal
justice. What is COMPAS, the algorithm in question? What are
ProPublica’s findings?
What three general properties a risk assessment algorithm should
satisfy? But what do we know in the end according to Theorem 2.1?
Reghavan also mentions that the theorem applies to decision-support
algorithms used in other areas such as employment, housing and finance.
Give your own specific example outside criminal justice.
Reconsidering broader contexts, Raghavan discusses two general directions
on pp. 95, 96. What are they?
What to hand in
Type your answers in two pages and submit them in paper.
C. Dwork, M. Hardt, T. Pitassi, O. Reingold and R. Zemel,
Fairness through awareness,
ITCS ’12: Proceedings of the 3rd Innovations in Theoretical Computer
Science Conference,
Cambridge, Mass., Jan. 8–10, 2012,
ACM, New York,
2012,
pp. 214–226.
J. Kleinberg, S. Mullainathan and M. Raghavan,
Inherent trade-offs in the fair determination of risk scores,
8th Innovations in Theoretical Computer Science Conference, ITCS 2017,
Berkeley, Jan. 9–11, 2017
(C. H. Papadimitriou, ed.),
LIPIcs. Leibniz Int. Proc. Inform., vol. 67,
Schloss Dagstuhl. Leibniz-Zent. Inform., Wadern, 2017,
Art. 43, 23 pp.
*Accessing this article requires subscription to the New York Times,
available for free to all Trinity students. If you do not have subscription,
read and follow the
Trinity College Libraries’ instructions
to register.
