It is a scary vision. However, the scary part seems to come not so much from the increasing data availability, but rather from the increased likelihood of discr

It is a scary vision. However, the scary part seems to come not so much from the increasing data availability, but rather from the increased likelihood of discriminatory uses of such data. Enforcing non-discriminatory uses of data sounds like a hard problem to tackle, especially if we realize how diverse is the space of possible missuses (e.g., in law enforcements, business, and politics) and that this problem affects both humans and computer systems. As hard as this problem sounds, the article provides a great motivation for research that defines computationally what "discriminatory" means, as well as for the research aiming to design systems (both human and computer) that are privacy-protecting and non-discriminatory.