A new weapon emerges on the modern battlefield. A weapon that radically changes the way wars can be fought. A weapon that challenges the notion of human control over technology and transforms the way decisions on the battlefield can and  will be made. Governed by sophisticated artificial intelligence, machines can observe, search for, loiter, engage, and destroy targets without human intervention. Once deployed, the Autonomous Weapons System (AWS) “decides” who lives to see  another day. This workshop focuses on Autonomous Weapons Systems at the frontier of automated killing and whether and how such systems can be regulated or used responsibly. The workshop presents experts with insights on the current use of AWS,  International Humanitarian Law, and the ethics of war. These scholars will constitute an interdisciplinary forum for discussions and principles, e.g. “meaningful human control”, transparency, distinction, and proportionality in decisions on life and death. Keywords. Artificial Intelligence, Ethics of war, International Humanitarian Law, Autonomous Weapons Systems, Automated killing, risk-free war, Just War Theory. 

Does risk-free war make any sense? The traditional Just War Theory, meaning “justified war”, rests, among other things, on the underlying logic of reciprocity among combatants. War entails the risk of losing your life, and since the combatant is under obligation to be obedient till death, the combatant has the right to self-defence and the right to kill the enemy’s combatants. The Rules of Armed Conflict and International Humanitarian Law stem from the Just War theory's fundamental principles. When combatants are removed from the battlefield and replaced by drones and other autonomous weapons systems, they are no longer in physical danger. They are also shielded from the psychological stress and moral damage of killing another human being. However, this, absurd as it might sound, disturbs the fundamental principles behind the ethics of war and the notion of war as a social activity. This presentation carves out some of the dilemmas that arise when war, on the one hand, is supposed to be something “humans do to each other” as a social activity and, on the other, about preserving the lives of both combatants and civilians. Dilemmas emerge when machines replace combatants, while Just War Theory remains the underlying logic and paradigm for the Rules of Armed Conflict.

The Israeli use of an AI system called Lavender to identify targets for bombing in Gaza might plausibly be held to mark the beginning of the age of “Minotaur Warfare”. Paul Scharre has argued that the future of war will be dominated by so-called “centaur warfighting”: teams of robots directed by human beings. The image of the centaur, a mythical beast with the body of a horse and the head and upper torso of a man, emphasises that a human being will be in charge of the manned-unmanned teams of the future. In a paper published in Parameters in 2023 [https://press.armywarcollege.edu/parameters/vol53/iss1/14], Adam Henschke and I argued that the future of war is more likely involve what we call Minotaur warfighting: teams of human beings directed by AI. The Minotaur was a mythological beast with the body of a man and the head of a bull. The image of the Minotaur highlights that the manned-unmanned teams of the future are more likely to have a “monstrous” head rather than a monstrous body. The use of Lavender in Gaza provides at least some evidence that we were right.

In this presentation I rehearse the considerations that led us to conclude that minotaur warfighting will outcompete centaur warfighting in the wars of the future and offer some initial thoughts on the ethics of putting AI in effective command of human warfighters. In particular, I argue that pressure to adopt minotaur warfighting is likely to move some military ethicists to reconsider their attitudes towards autonomous weapon systems. 

A March 2021 United Nations report argued that the Kargu-2, a one-way attack drone, has been used to strike militias in Libya autonomously. In the war in Ukraine, both sides have used similar types of drones that appear to have the latent technical capability to identify, track, select, and strike targets autonomously. Israel’s use of AI systems to generate target lists for attacks in the Gaza strip are well-documented. These examples underscore a trend toward the deployment of autonomous and AI technologies in warfare. While often discussed under the umbrella term of autonomous weapon systems (AWS), most current systems used in targeting decision-making appear to be operated with humans ‘in the loop’ to authorise attacks. But the quality of control and agency that humans can exercise is already compromised due to complex tasks and expected decision-making speed. I examine how this diminishing quality of human control results from a governance gap of AI in the military domain. In the absence of top-down governance, I argue that the use of autonomous and AI technologies in weapon systems makes norms. Practices of design, of training personnel for, and of using such weapon systems in targeting decision-making shape a social norm, defining what is considered an “appropriate” level of human control. This emerging norm accepts a diminished form of human control, thereby undercutting human agency in warfare. While there are a growing number of governance initiatives on AI in the military domain, these initiatives do not necessarily scrutinise this emerging norm.  

Systematic killing has long been associated with some of the darkest episodes in human history. Increasingly, however, it is framed as a desirable outcome in war, particularly in the context of military AI and lethal autonomy. Autonomous weapons systems, defenders argue, will not only surpass humans militarily, but morally, enabling a more precise and dispassionate mode of violence, free of the emotion and uncertainty that too often weakens compliance with the rules and standards of war. We contest this framing. Drawing on the history of systematic killing, we argue that lethal autonomous weapons systems reproduce, and in some cases, intensify the moral challenges of the past. Autonomous violence incentivises a moral devaluation of those targeted and erodes the moral agency of those who kill. Both outcomes imperil essential restraints on the use of military force. 

The existing legal framework for military operations during armed conflicts, international humanitarian law (IHL), contains a set of obligations aimed at protecting civilians and mitigating the suffering of those exposed to armed conflicts. While there is broad agreement that the technology-neutral targeting rules of IHL apply to autonomous- and AI-enabled weapons systems, it is less clear what is needed to ensure compliance with this body of law in practice. Increased autonomy and reliance on machine learning and other AI-disciplines in the critical targeting functions of weapons systems inevitably create  new types of operational and legal risks which will most likely multiply as machine and conflict complexity increase. This paper argues that the highly context-specific nature of IHL obligations calls for a risk-based approach to governance and comprehensive and rigorous risk mitigation measures tailored to meet the specific needs of each discrete system and circumstances of use, something that neither can nor should be achieved through the enactment of new treaty law alone. On this basis, the paper proceeds to identify the latest examples of international and regional cooperation around the creation and practical implementation of broader responsible AI frameworks and discuss their potential contribution to enhancement of IHL compliance. 

Technology has always played a vital role in the Western conception of itself. It has firstly functioned as an identifier: technological mastery was and continues to be drawn upon as a marker of “civilized peoples” and “civilized warfare”. Today, we see this expressed through the Western emphasis on technologically-enabled battlefield precision, and “humanely” delivered violence. Technology has also functioned as safeguard: technological advancement has been valued by Western actors as a way to militarily contest and dominate others, including those deemed “uncivilized”. Though linked through a shared assumption of civilizational superiority, these twin features of technology have historically produced a tension, between violence with limits and violence without. In this article, I evaluate both features of technology in relation to the ongoing pursuit of military AI and autonomous weapons. Imaginaries of Western civilization, I argue, infuse the framing of military AI as a humanizer of war, while also giving license to its unrestrained development and use.