Jonne Maas, PhD candidate at TU Delft involved with the HumanE AI project.

Abstract: This talk is based on a project for Humane-AI net, in which we critically reflected on the design of an AI system by investigating the role of the designer. Designers make choices during the design of the system. These choices have an influence on the system, hence design choices have important epistemic and moral value. Analysing these choices and their effective consequences contributes to an overall understanding of the designer’s situated knowledge embedded in a system. The reflection is concerned with questions like how do designer’s interpretations affect the output of a system, and hence touches upon the power dynamics underlying the development of an AI system. We claim that design choices ultimately have epistemic and moral limitations, highlighting the special role of an AI designer within society. To ground our abstract claims in real life, we followed the design of an agent-based social simulation aimed at modelling how economic inequality affects policy-making.