Contextualizing Coordination in the Dutch AI Oversight Network
Maurus Enbergs (TU Delft - Technology, Policy and Management)
Roel I.J. Dobbe (TU Delft - Technology, Policy and Management)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The Oversight of Artificial Intelligence is emerging as a complex challenge. To tackle this complexity, regulators are increasingly required to collaborate on AI oversight in networks. Yet, little is known about how the coordination of AI oversight must be organized in these arrangements. This paper identifies factors that influence the coordination of AI oversight in regulatory networks, focusing on the Dutch Samenwerkingsplatform Digitale Toezichthouders (SDT). We conducted semi–structured expert interviews with members of AP, DCA (Directie Coördinatie Algoritmes), who are tasked with facilitating collaboration and coordination of AI Oversight in the SDT network. We identify overarching themes that influence coordination in three interlocking core AI oversight activities: risk monitoring; joint evaluation of AI systems; and joint advice for legal compliance. Our findings show that AI oversight has wicked characteristics whose resolution is dependent on the critical functions of knowledge integration. To strengthen knowledge integration, we recommend that regulators work towards system-level assessment methods for AI systems that readily allow them to combine their different perspectives. We believe that such assessment practices should embrace the principles of knowledge co-production. Our descriptive approach helps inform coordination of AI oversight in the Netherlands and elsewhere.
Files
File under embargo until 06-03-2027