Fairness, Accountability and Transparency: Friction and Reconciliation
Philosophy and Public Affairs, 6 June 2024 | Faculteitskamer Oude Turfmarkt 145 | Speaker: Prof. dr. Linnet Taylor (TiU) | Commentator: TBA | Contact: Tijn Smits - t.m.smits@uva.nl
Event details from: Fairness, Accountability and Transparency: Friction and Reconciliation
- Date
- 6 June 2024

Contact: Tijn Smits - t.m.smits@uva.nl
Abstract. The past few years have seen growing recognition that machine learning raises novel challenges for ensuring non-discrimination, due process, and understandability in decision-making. They have also witnessed an unprecedented increase in the complexity of machine learning models, which challenges efforts to address the concerns above. This chapter will focus on identifying these concerns (e.g. algorithmic bias and discrimination) and on the challenges met, but also posed by, their proposed solutions. The work of addressing fairness, accountability and transparency in machine learning systems takes place at the intersection of ethics, political philosophy, law and regulation, and social justice. There is perhaps no other object of multidisciplinary research that has generated so much contemporary conflict. In order to understand why, we will outline both the history of the debates that have arisen over fairness in relation to machine learning, and the worldviews that are coming into conflict. We will use the concept of levels of abstraction (Floridi, 2008) to consider the different conceptual levels these debates address, and to analyse why they are not easily reconciled. We will then outline possible ways to foster generative, rather than destructive, friction around the concept of fairness, and how these conflicts can speak to each other in a useful way.
Bio. Dr. Linnet Taylor is Professor of International Data Governance at the Tilburg Institute for Law, Technology, and Society (TILT), where she leads the ERC-funded Global Data Justice project. Her research at TILT broadly focuses on how new sources of digital data are impacting governance, research on human and economic development, and political representation.