HMI DAIS 05 - Barbara Kiviat, Stanford University
HMI DAIS 05 - Barbara Kiviat, Stanford University
Public online seminar, 9am 20 August 2020 AEST
Barbara Kiviat of Stanford University gave the fifth HMI Data, AI and Society public seminar.
Barbara Kiviat is Assistant Professor in the Department of Sociology at Stanford University. She is an economic sociologist who studies how moral beliefs and other cultural understandings shape markets and justify the inequalities they produce. She is particularly interested in how normative ideas influence the pricing and allocation of socially important resources, such as insurance, credit, and jobs. Her current project considers how these dynamics play out when corporations use massive amounts of personal data to decide what to offer to individual consumers. She mostly uses qualitative methods, but also works with survey data and vignette experiments.
Kiviat’s research has received awards or funding from the American Sociological Association, the National Science Foundation, the Society for the Advancement of Socio-Economics, the Washington Center for Equitable Growth, the Edmond J. Safra Center for Ethics, and other groups. Her work has been published in American Sociological Review, Socio-Economic Review, Socius, and Social Service Review.
Title: ‘The Moral Legibility of Narrative and Case Comparison’
Abstract: In evaluating how algorithmic predictions allocate benefits and burdens, on-lookers often turn to storytelling, causal theorizing, and other narrative ways of knowing. In this talk, I argue that one reason narrative factors so prominently in debates about algorithmic fairness is that stories provide information that algorithms—rooted in the competing logic of case comparison—do not. Whether we understand people as actors in unfolding series of events or as cases with particular attributes determines which aspects of individuals and situations are legible, and, as a result, the types of moral judgments we can make. I start with an example: the decades-long public policy debate about whether it is fair for U.S. car insurers to use consumer credit scores to set prices. Here, policymakers insisted on narrative (as well as actuarial) accounts of credit scores in order to assess whether motorists were getting the insurance rates they deserved. Building on this example, I outline the ways narrative and case comparison organize cognition differently. Each mode of sensemaking brings different orientations to time, mental states, affect, social relations, causality, and more. In the final part of the talk, I show how some notions of (in)justice are easier to adjudicate when we think narratively, while others are more visible when we focus on comparison across cases.
HMI Dais recordings can be viewed here.