Philosophical Grounds for Regulating the Lawful Operation of Autonomous Vehicles

Abstract

The predictability of algorithms will allow us, for the first time in history, to determine in advance how autonomous vehicles (AVs) will respond to unavoidable traffic accidents. This article delves into consequentialism, including act and rule utilitarianism, and deontology to explore how algorithms should make decisions in various crash situations. As a solution, it advocates for adopting a hybrid structure that incorporates elements of both theories and leverages flexible machine learning algorithms. Based on this structure, the Article proposes mandatory ethical settings that prioritize minimizing overall damage, avoid discrimination based on sex, wealth, age, race, and similar factors, and ensure explainability in life-and-death decisions. Such regulation is necessary, as emerging laws on AI, like the European Union Artificial Intelligence Act, do not address specific actions for AI systems in the event of inevitable accidents. Furthermore, the Article urges early regulation before the widespread adoption of AVs to prevent occupant-protectionist AVs from gaining competitive advantage and becoming the norm in the market. For individuals harmed under these ethical settings who seek full recovery from liable parties, the practical difficulties associated with the burden of proof should be reconsidered following the legal amendments. © 2025 Informa UK Limited, trading as Taylor & Francis Group.