Autonomous weapons are the latest military technology to receive major attention outside expert circles. Like drones before them, they are heralded as revolutionary. And, as with drones, there are major political and ethical concerns. As there is widespread agreement that autonomous weapons systems (AWS) will be developed and are likely to be used on future battlefields — unless this development is deliberately halted — it is of the utmost importance that the public and policymakers have an informed debate on what this means for warfare, the military, and society more broadly. Academia has an important role to play in this process.
Autonomous Weapons Systems: Law, Ethics, Politics contributes to this debate, although its target audience is academics rather than a broader public. The book addresses the emergence of autonomous weapons from a legal point of view, aiming to clarify key areas and develop questions for future research. It consists of 15 contributions that are organized into five sections: (1) “Meanings of autonomy and human cognition under automation”; (2) “Autonomous weapons systems and human dignity”; (3) “Risk, transparency and legal compliance in the regulation of autonomous weapons systems”; (4) “New frameworks for collective responsibility”; and (5) “New frameworks for individual responsibility.”
The majority of the authors are law professors, but the book also includes contributions from experts in the humanities and social sciences, as well as a professor of robotics and a professor of engineering. Most of the contributors are European and/or based at European institutions, which is a positive change from the usually American-centric AWS debate. Although the authors do not specifically address European developments, this is still a good sign as it shows that there is interest in and work done on AWS in Europe.
An early limitation of this volume is the lack of a common definition of automation. Christof Heyns, former United Nations Special Rapporteur on extrajudicial, summary, or arbitrary executions, notes in his essay, “Full machine autonomy has not yet been used against human targets, but the point has been reached where that possibility has become very real” (p. 5). In the public debate, the term “autonomous weapons” usually refers to the automation or the autonomous decision to use force, specifically, force directed against individuals. As a reader, it would have helped if the authors had agreed on a common definition of automation and autonomy — even if the authors in their respective chapters had chosen to explain why, and how, their definition differs from the common one. This would have saved the reader from attempts to define autonomy in each chapter.
Weapons autonomy is becoming a heatedly debated topic, yet autonomy or automation up to the point of the use of force often receives less attention than the automation of the use of force. Many people have an intuitive dislike of the idea of AWS that are able to make decisions over life and death. Faced with this moral consideration, people may intuitively agree with Heyns, who argues that “a machine, which is bloodless and without morality or mortality cannot do justice to the gravity of the decision whether to use force in a particular case, even if it may be more accurate than humans” (p. 11). Whether a machine can do justice to the gravity of the decision to kill is one of the core questions addressed in this volume.
The German ethics professor Dieter Birnbacher, for example, addresses the question of whether AWS are a threat to human dignity. It is often stated that robotic weapons “dehumanize” warfare and violate human dignity. Birnbacher shows convincingly that this statement is problematic insofar as human dignity lies with the individual, not with humankind. Crucially, Birnbacher asks, “can it really be said that AWS humiliate those who are chosen by them as targets or taken in the bargain as incidental victims?” (p. 115). To what extent does it matter for the victim whether he or she is killed by a missile, by a bullet, in a hand fight, or by an AWS? As Birnbacher puts it, “Of course, machines cannot comprehend the value of human life. But why should this make a difference to their victim, if alternatively, they are threatened with being wounded or killed by manned weapons such as bombers? For the victims whose dignity is at stake, it is a matter of indifference whether the threat they are exposed to comes from manned or unmanned weapons, provided all other parameters of the situation are equal” (p. 120).
Noel Sharkey, a well-known activist against AWS and emeritus professor of artificial intelligence and robotics, warns in his contribution that AWS are not likely to fully comply with international humanitarian law (IHL) (p. 24). While this view is controversial, he asks a pertinent question: “If our thinking, our strategies and our funding are directed towards developing AWS, and it turns out that making them IHL compliant is not as successful as was hoped for, what will we do with this weapons technology?” (p. 25). Sharkey underlines an important fact — namely, that human involvement in the decision to use force does not necessarily mean meaningful human control. “To say that there is a human in the control loop does not clarify the degree of human involvement” (p. 26). However, as I previously discussed, many people are not interested in the automation of weapon systems, as long as the decision about when and where to use force lies with a human. Yet this approach can be treacherous as it is questionable how much meaningful control the human still has over the decision if hr or she is unable to follow and understand the logic that resulted in a building or person being singled out as target by an AWS.
From a policy perspective, a significant chapter is Guglielmo Tamburrini’s discussion “On banning autonomous weapons systems: from deontological to wide consequentialist reasons.” In it, the robotics professor from the University of Naples notes the misunderstandings that exist between experts and the public, which hampers and complicates the public debate on AWS. This is particularly prevalent with regard to developments’ time horizons. “Experts speaking to restricted circles of peers may abstain from signalling whether they are speaking from the long-term or short-term perspective of a research programme…. In public statements about their work, however, experts can no longer count on the shared background of tacit knowledge that shapes communication styles within communities of experts” (p. 135). This leads to misunderstandings about the likelihood and time horizons of developments such as singularity, full autonomy, and the rise of “killer robots.” As Pablo Kalmanovitz points out in the following chapter, it is problematic if the public is unable to understand the issues pertaining to AWS: “Ultimately, …public scrutiny may be the only available effective push for IHL compliance and accountability” (p. 146).
Other contributions that deserve attention are those by Geoffrey S. Corn, as well as Eliav Lieblich and Eyal Benvenisti’s chapter, which may have been better placed at the start of the book, as it describes well “the circular debate on AWS” (p. 251), which confronts technological optimist and pessimist, boiling down to a question of faith in technological progress.
Overall, Autonomous Weapons Systems is a useful contribution to the ongoing debate about AWS and lays a foundation for future discussions. However, as it suffers from excessive jargon and is at times unnecessarily complicated, the book is best suited for an academic audience rather than policymakers. Individual chapters can be assigned in teaching at the graduate level, in particular in the area of legal studies.
