CYIL vol. 9 (2018)

CYIL 9 ȍ2018Ȏ AUTONOMOUSWEAPONS SYSTEMS: LAW, ETHICS, POLICY with a text that follows the historical as well as modern approach to machine autonomy, learning and behaviour-based robotics. These contributions are more descriptive and technical than others in the volume. However, since autonomy is the decisive feature of the systems that are in the centre of attention here, it is paramount that we understand what machine autonomy is, what it entails and how it differs from human autonomy. All authors in this part agree that increasingly autonomous systems should strive for the ideal partnership that allows the human and the machine to operate together in a way that utilizes the strengths of both, but the human should always maintain sufficient control over the machine. Several academics have recently asked a question whether AWS, even if they could respect the rules of international humanitarian law (“IHL”), would still violate human dignity. This question is addressed in the third part of the book in a contribution by Dieter Birnbacher . He presents human dignity as an overarching and open-ended concept, which, in his view, carries a danger of overstretching it and interpreting it improperly. Human dignity should be approached as a concept applicable only to individuals (not humanity as a whole). It should imply a number of basic human rights, the breach of which carries a particularly strong moral reproach. According to Birnbacher the greatest threat to human dignity in relation to AWS lies in the risk of violating the right to a minimum quality of life and relief from suffering and the right not to be treated merely as a means to other people’s ends with severe risk of harm. Among features of AWS that cause this threat Birnbacher lists the unpredictability of their responses, limited capacity to observe the rules of proportionality and discrimination and their deployment in asymmetrical warfare. In any case, he stresses that from the point of view of the individual whose dignity is threatened it does not really matter whether the danger comes from an AWS or a manned aircraft. Therefore, AWS are not necessarily inherently incompatible with human dignity. Their use, however, could be. The next three sections of the book focus on the issues surrounding the accountability for the use of AWS. The so called “accountability gap” that the AWS create, according to the majority of experts familiar with the topic, is seen as the most pressing problem. As such, it gets most attention in the collection. The contributions discuss at length transparency and information sharing with respect to military technologies, the rules for conducting reviews of new weapons, individual criminal responsibility and other frameworks for responsibility for the acts of AWS. The main issue that resonates throughout all the texts in these sections is the unpredictability of AWS. Given their autonomous behaviour, the various humans involved in the use of AWS could lack sufficient control and foresight in relation to the actions of AWS. Several authors therefore mention that although there would always be humans involved at some level in the design, programming, manufacturing, deployment and decision to activate AWS, the existing legal norms may be inadequate to effectively attribute individual criminal responsibility to them. According to Hin-Yan Liu , this could lead either to impunity for the outcomes caused by AWS or to the scapegoating of the humans involved due to over- extending the limits of responsibility for the actions that could be far beyond their possible influence. Particularly interesting are two contributions that draw inspiration from different branches of law to establish accountability for the actions of AWS. Nehal Bhuta and Stavros-Evdokimos Pantazopoulos turn their attention to international environmental law and propose to adopt an approach to the design, testing, verification and authorization of the use of AWS similar to the one taken in relation to the transboundary harm resulting from

423