• Ascriptive Algorithms-Exploring the Classificatory Influence of Predictive AI Technologies

Systems of classification surround our daily lives. But classification comes with consequences: scholars have long argued for the ability of classification schemes to illuminate, validate, and order. In this paper, I examine the classificatory influence of predictive AI technologies. I argue both that the use of predictive AI tools holds normative classificatory consequences, and that extant work in the philosophy of social science can help us to understand them. To this end, I make three main claims. First, I suggest that predictive AI tools hold the power to classify, and that this power shapes our own self-conceptions. Second, I suggest that AI tools have the power to illuminate new features of categories thanks to an inherent proxy-searching feature of their system design, and that this serves as evidence for social categories as both constructed and real. Finally, I claim that AI tools entrench the categories they illuminate, which thereby further solidifies the reality of those categories. I unpack and assess the implications of each of these observations, and conclude by suggesting future research directions in light of them. Given that predictive AI technologies are becoming increasingly ubiquitous, there is an urgent need to explore their ethical impacts. This paper, in interrogating an as-of-yet underrecognized feature of AI’s influence, thereby serves to contribute to a more robust understanding of AI and our social world.