Artificial intelligence (AI) seems to change the way insurers assess risks and set prices: insurers can now analyse large amounts of new data types — ranging from Internet of Things (IoT) devices to social media. This may allow insurers to estimate risks in greater detail. In this dissertation, I address the central research question: To what extent do insurers introduce discrimination-related risks when they use AI systems for underwriting? I organise the PhD thesis around four topics: trends in how insurers apply AI and the possible risks of these practices, the legal obstacle to testing for bias under the General Data Protection Regulation (GDPR), the effectiveness of a de-biasing-exception in the EU AI Act, and I evaluate, using a first survey, how the public might evaluate these practices.
Marvin van Bekkum is a PhD Candidate at the Interdisciplinary Hub for Digitalization and Society (iHub) and the Institute for Computing and Information Sciences (iCIS). His PhD supervisors are Prof. Dr. F.J. (Frederik) Zuiderveen Borgesius and Prof. Dr. T.M. (Tom) Heskes. Marvin studied Civil Law and Cyber Security at Radboud University (LLB, BSc, LLM, MSc). He has a broad interest in the crossroads between law and technology.