The wealth of data available in the modern age has enabled the use of machine learning methods and other data science methods in a range of new areas. Current applications include ranking items in social media feeds, optimizing advertisements, and surveillance and predictive policing by government and law enforcement. This talk will discuss some of the potential ethical and privacy issues associated with the widespread use of machine earning algorithms. Most suffer from a lack of transparency in their design and operation. Mass social engineering is feasible through the use of individualized messages crafted by adaptive algorithms. Subtle manipulation would be very difficult to detect by individuals but can have significant social impact. In addition, biases in input datasets used for training algorithms treated as impartial can systematize discrimination against certain populations. Faced with these challenges, some potential avenues for ameliorating these problems will be discussed, both in terms of policy and technology. As a community, we need to better understand and monitor the role of these methods in society in order to ensure that we build and support systems that are resistant to misuse.
Jennifer Helsby (@redshiftzero) is a researcher and scientist. She is currently a Data Science for Social Good fellow at the University of Chicago where she works on the application of data analysis and machine learning methods to problems with positive social impact. She also is a co-organizer of Cryptoparty Chicago, which teaches people about privacy issues and digital security practices. Before that, she completed a Ph.D. in Astrophysics at the Kavli Institute for Cosmological Physics at the University of Chicago where she did theoretical and computational work studying the large scale distribution of galaxies in the universe.