The Ethics of Automated Vehicles

Abstract

My thesis is about the morality of automated vehicle (AV) decisions. What is the relevance of the trolley problem to AV decisions? What is the morally right method for AVs to distribute harms or risks of harm between the parties in collisions? Who is morally responsible for harm caused in collisions? How does the risk of harm to road-users trade-off against the AV’s prudential goal of getting to its destination in reasonable time? What is the morally right amount of caution for AVs to exercise when uncertain about the classification of proximate objects? I will answer these questions by developing a deontological account of permissible killing for AVs and a deontological account of permissible risk-imposition. I also argue against some of the rival answers to these questions; and in doing so, attempt to show that my view does a better job than its rivals at capturing our considered moral judgements.

The view I defend holds that the AV is morally permitted to kill or harm a road-user in an unavoidable collision if, and only if, and because, its passenger is morally permitted to kill or harm that road-user in self-defence. In normal driving, the AV is morally required to moderate its speed so that it can safely negotiate modally close but improbable what if cases such as children running out into the road. Here is the plan. Chapter 1 is a literature review. Chapter 2 defends my use of trolley cases. Chapters 3 and 4 argue against two rival non-consequentialist views, Filippo Santoni de Sio’s (2017) legal-philosophical view and Derek Leben’s (2017) Rawlsian view. Chapter 5 presents my deontological theory. Chapter 6 presents my account of risk-imposition, and develops an illustrative formal decision-procedure for a simple mundane road-traffic scenario modelled as a Markov Decision Process.

Date of Award	23 Jun 2020
Original language	English
Awarding Institution	The University of Bristol
Supervisor	Richard G Pettigrew (Supervisor) & Brad Hooker (Supervisor)