Automated Vehicles and the Ethics of Classification

Abstract

AbstractThis paper articulates the moral significance of object classification for automated vehicle decision-making. The first part of the paper examines how classification errors and uncertainty about the classification of proximate objects relate to the permissibility of an automated vehicle’s actions. It is argued that moral permissions for automated vehicles are indexed to their epistemically justified predictions about the classification of proximate objects. The second part of the paper develops an account of the morally right way to navigate uncertainty about the classification of proximate objects in mundane road traffic situations. In particular, it is argued that an automated vehicle acts wrongly if, given its evidence, it could easily be the case that the vehicle’s actions injure a road-user. The paper concludes by developing an account of how to balance the competing values of time-efficiency and road-user safety in mundane road-traffic scenarios that involve uncertainty about the classification of proximate objects.