Colloquium: Moral Agency and Machine Learning

18 May 2022

Colloquium by Prof. Dr. Geoffrey Bowker (University of California Irvine, USA)

Abstract: Moral Agency and Machine Learning

It is often claimed that machine learning does not use classification systems assigning people to ‘protected’ categories and that therefor it is unbiased. However, its effects often include clear value judgements and political bias. I present a framework for thinking through the question of whose decisions get folded into machine learning and for developing an analytic framework for their exploration. I examine especially the claim that there is no role for classification in machine learning: to the contrary, the political act of classification recurs at each of the many levels at which it operates – from deciding what is ‘good’ data to input, through the reification of categories (and its political implications) through our understanding of users. I briefly compare policy frameworks around machine learning in the United States and Europe, and suggest a broadening of these based around empirical studies of the practice of machine learning, arguing that much policy is deficient precisely because it blackboxes practice.

About the speaker: Geoffrey C. Bowker is Emeritus Donald Bren Chair at the School of Information and Computer Sciences, University of California at Irvine. Recent positions include Professor of and Senior Scholar in Cyberscholarship at the University of Pittsburgh iSchool and Executive Director, Center for Science, Technology and Society, Santa Clara. Together with Leigh Star he wrote Sorting Things Out: Classification and its Consequences; his most recent books are Memory Practices in the Sciences and (with Stefan Timmermans, Adele Clarke and Ellen Balka) the edited collection: Boundary Objects and Beyond: Working with Leigh Star. He is currently working on time and computing under the rubric ‘life at the femtosecond’.

About the Lecture Series: This semester’s edition of Taming the Machines looks at the data and design practices that surround the development of information technologies and artificial intelligence. What are the challenges of creating practices that enrich human life and society? Which practices best harvest the positive potential of data? How should we design our machines and algorithms so that they do not endanger our humanity or agency? How do societal imaginaries shape information technologies? Which positive visions for sociotechnical futures do we need?
To answer these questions, the public lecture series invites distinguished researchers to talk about the data practices and design cultures that surround the development of information technologies and artificial intelligence.

Date and time: Wednesday, 1 June 2022, 18:15, Zoom Meeting.

The lecture is part of the public ring lecture series "Taming the Machines". Please visit https://www.inf.uni-hamburg.de/en/inst/ab/eit/taming-the-machines/summer22.html to register for the webinar.