Algorithmic Lunacy and What to Do About It
Women are less likely to be shown ads for high-paying jobs. Search results for "human rights" depend on what US state you search from. A health app warns the disabled that they "lack vitality." Machine learning labels Auschwitz photos "sport." Algorithmically-generated content is responsible for a growing list of fiascoes, causing alarm about the future of these systems. As the scandals mount, platforms are responding with public relations and friendlier algorithmic controls (such as Facebook’s “See First” feature). We debate the state of play in the domain of algorithmic content, first assessing the disasters, then evaluating both user and platform strategies that aim to tame the black box.
Presenters

Christian Sandvig
Assoc Professor
University of Michigan
Christian Sandvig is Steelcase Research Professor and Associate Professor in Communication and Information at the University of Michigan. He is also a Faculty Associate of the Berkman Center for In...
Show the rest
Dawn Nafus
Sr Research Scientist
Intel Corporation
Dawn Nafus is an anthropologist at Intel Labs, where she informs new technology design and strategy. She holds a PhD from University of Cambridge, and is interested in how people experience time, ...
Show the rest