A January@GSAS mini-course on ethical and computational approaches to more fair and private decision-making with big data.
Updated: 22 Jan 2020
In the ‘Wild West’ of applied big data analytics, human experiences are increasingly driven by machine predictions that can unexpectedly violate fairness and privacy at scale.
Unless we bridge the gap between industry and research, and the gap between policy and technology, unfair and non-private practices will continue to persist in the growing number of decision-making processes that are driven by data. For this reason, this is not intended as a computer science, statistics, ethics, or law course. In this mini-course, we will explore new ethical and computational approaches to more fair and private data usage through the interdisciplinary lens of five real-world case studies. Read more about the case method.
Tuesday, 21 Jan
This introductory seminar will introduce the concepts in computer science and applied data science that underpin the case scenarios discussed in the remainder of the course.
Required Reading
Recommended Readings
Tuesday, 21 Jan
In this seminar, students will learn and discuss the legal and ethical framework that governs fairness and privacy in the information age.
Required Reading
Recommended Reading
Wednesday, 22 Jan
Big data and big computation have enabled online advertising platforms to maximize revenues by serving the right ads to the right consumers, with increasingly specific targeting parameters offered to advertisers. Kicking off with a case scenario, students will explore the impact to fairness and privacy of services that are free to consumers because the consumer is the product.
Required Reading
Recommended Reading
Wednesday, 22 Jan
In the United States, and in many other legal jurisdictions, health data is subject to special privacy and use protections. However, analysis of micro and macro health data presents major opportunities to benefit individuals, public health, and medical research. Nonetheless, “black box” algorithms are controversial when applied to a context so intrinsically human as human health. Through the case method, students will explore ethical and technical approaches to find a balance between costs and benefits of algorithmic healthcare.
Required Reading
Recommended Readings
Thursday, 23 Jan
Government is unique among users and brokers of data, with the vast administrative data it collects on individuals, its ability to compel action, and the monopoly on state violence. Through a representative case scenario, students will explore when government is required to act in a fair and private manner.
Required Reading
Recommended Reading
Thursday, 23 Jan
Algorithmic decision-making was introduced in the judicial system to remove bias from how an individual experiences the courts. With a case study, students will consider whether there are cases where algorithmic decision-making formalizes and scales historical bias instead of removing bias from the system, and explore practical remedies.
Required Reading
Recommended Readings
Friday, 24 Jan
Risk assessments play a significant role in a bank’s decision whether or not to extend credit to a consumer. The FinTech industry is considering ways to make the algorithms more accurate (any by their logic more fair), but the data that even non-traditional risk assessments consider remains controversial. A case on the use of algorithms in lending decisions will prompt students to explore the ways machines promote - or prohibit - fair lending practices.
Required Reading
Recommended Reading
Friday, 24 Jan
In this closing seminar, students will synthesize lessons learned from the application of the legislative and ethical framework to real-world case studies, and discuss how these can be applied to promote fairness and privacy in practice. What is the law? What should the law be? How can data-intensive applications and individual rights to privacy and fairness coexist?
Recommended Reading