Kars Alfrink is a PhD candidate at Delft University of Technology’s faculty of Industrial Design Engineering. His main research interest is ensuring AI systems respect the human right to autonomy by making them contestable by design. His work uses forms of urban AI as cases which connects him closely to the BOLD Cities agenda.
Can you tell us about your current research?
I focus on AI systems used by local governments in the delivery of public services. For example, so-called “scan cars” used for parking enforcement, and smart charge points for electric vehicles. I collaborate on these projects with the Responsible Sensing Lab at the Amsterdam Institute for Advanced Metropolitan Solutions (AMS Institute).
I aim to develop new knowledge for the design of mechanisms that enable people to meaningfully contest individual automated decisions made by AI systems; and collectively contest the design and development of AI systems, particularly as it pertains to datasets and models.
My PhD research team consists of Neelke Doorn, Professor of Ethics of Water Engineering at the Department of Values, Technology and Innovation; Gerd Kortuem, Professor of Internet of Things at the Department of Sustainable Design Engineering; and Ianus Keller; Lecturer at the Department of Human Centred Design.
Why do you think an interdisciplinary approach is needed for smart city related research/this research project?
We need a broad variety of disciplines to create a complete view on this multifaceted topic of AI and contestability. I find myself straddling the fields of design, computer science, philosophy, policy, law and the social sciences. The only way to make sense of this complexity is through interdisciplinary collaboration.
Some argue that contestability creates even more complexity and friction. I would say the administrative logic of efficiency has for a long time dominated the drive towards automated decision-making in public governance and elsewhere. Contestability is a way to give more prominence to the democratic values of inclusivity and pluralism. I find it very important in this project to collect a diversity of points of view and to remain open to possible contradictions. I believe contestability is a way to leverage disagreements for the continuing improvement of AI systems.
What was your personal motivation to become affiliated with the Leiden-Delft-Erasmus Centre for BOLD Cities?
The BOLD Cities research operates on a scale much larger than what I can achieve by myself in a PhD project. The centre also helps me share these ideas and results with those organizations for which they might be useful. I find the connection between citizens and government very important for the democratic legitimacy of AI.