UAS as moral agents: Dilemmas and solutions

Unmanned aircraft systems (UAS) have the potential to bring many benefits, but first they must be accepted by society. Such acceptance will come only if UAS perform their missions safely. Furthermore, in dangerous situations where safe outcomes may not be possible, UAS decision making must be just, equitable, legal and moral. UAS flight control algorithms will inevitably be required to make decisions with safety implications, and this means the UAS will be required to make decisions with moral implications. This becomes particularly important in safety scenarios where a choice between different degrees of harm is unavoidable. In such scenarios, UAS algorithms must make hard ethical choices. And when harm occurs, even subtle choices made by the flight control algorithms will be scrutinized. It is therefore essential that software engineers, policymakers, regulators, and operators have a clear understanding of societal expectations and values when designing software that will drive the decision-making processes of increasingly automated systems. This paper explores the technical, legal, societal, policy and moral aspects of this problem and candidate research methods on which development of a moral UAS Universal Value Schema might, in the future, be based.