If the enthusiasts are to be believed, we are standing at the cusp of a momentous upheaval in the character of warfare, brought about by the large-scale infusion of robotics into the armed forces. In the future, many proponents of the “robotics revolution” argue, a variety of military functions—from simple logistical tasks to the application of lethal force—will be performed by machines acting more or less autonomously, without any direct interference from human operators. Human-rights groups and other critics agree that such a change might be in the offing—and argue that it must be urgently reined in before machines devoid of any sense of moral reasoning are given the capacity to make unaccountable, independent decisions that lead to the taking of human lives.

While it is important to consider the implications of autonomous weapon systems (AWS) from a variety of perspectives, moral and legal standpoints—forcefully articulated by activists often driven more by “righteous indignation” than by a penchant for detached analysis—currently dominate the debate. Meanwhile, their strategic consequences generally remain underexplored and are, at best, alluded to nebulously.

But what might the effect of such autonomous weapon systems pose for military stability? Will these systems increase the likelihood that militarized rivalries lead to destabilizing arms competitions? Will they undermine the purpose of deterrence by creating incentives for actors involved in a crisis to strike first? And will they contribute to situations in which hostilities, once joined, quickly get out of hand?

If the impact of AWS on arms-race stability, crisis stability and the prospects of escalation is unproblematic, then they would tend to support military rationales for developing and procuring such systems. If, on the other hand, their impact is found to be deleterious, this should provide decision makers with an added incentive to carefully weigh the potential advantages of such systems against their negative consequences for national security.

Autonomous Warfare: A Likely Prospect

Military forces that rely on armed robots to select and destroy certain types of targets without human intervention are no longer the stuff of science fiction. In fact, swarming anti-ship missiles that acquire and attack targets based on pre-launch input, but without any direct human involvement—such as the Soviet Union’s P-700 Granit—have been in service for decades. Offensive weapons that have been described as acting autonomously—such as the UK’s Brimstone anti-tank missile and Norway’s Joint Strike Missile—are also being fielded by the armed forces of Western nations. And while governments deny that they are working on armed platforms that will apply force without direct human oversight, sophisticated strike systems that incorporate significant features of autonomy are, in fact, being developed in several countries.

In the United States, the X-47B unmanned combat air system (UCAS) has been a definite step in this direction, even though the Navy is dodging the issue of autonomous deep strike for the time being. The UK’s Taranis is now said to be “merely” semi-autonomous, while the nEUROn developed by France, Greece, Italy, Spain, Sweden and Switzerland is explicitly designed to demonstrate an autonomous air-to-ground capability, as appears to be case with Russia’s MiG Skat. While little is known about China’s Sharp Sword, it is unlikely to be far behind its competitors in conceptual terms.

In light of these developments, a future in which armed platforms execute some missions and attack some types of targets autonomously is certainly imaginable—perhaps even likely. What is beginning to take shape in the air will also be true of other operational environments. What, then, will be the strategic consequences of the inclusion of such systems into advanced force structures, as far as military stability between states and potentially even between states and nonstate actors is concerned?

Wrecking the Balance

From swords made of iron, to the breech-loading rifle, to the Dreadnought-type battleship, to multiple independently targetable reentry vehicles (MIRVs), the introduction of novel military capabilities that provide one actor with a substantial edge has usually forced others to follow suit. The pressure for competitive force modernization, which is created by the adoption of new technologies and new doctrines for employing them, has sometimes resulted in self-sustaining arms races. These are commonly assumed to make war more likely, although international-relations research now tends towards more complex and less gratifying narratives. It is nonetheless important to ask whether the adoption of AWS could result in a destabilizing arms competition, and there are reasons to believe that it might.

First, if AWS should prove much more capable than human-controlled systems in important areas of warfare, advanced armed forces the world around may find it difficult to exclude them from their force structures for very long, even if they have serious misgivings about them. Pressure for adoption could result from the desire to gain an important edge over a competitor, from the fear of being preempted by others, or reactively, once the systems have proved themselves superior on the battlefield. This may well be the case in activities such as air-to-air combat, in which extremely rapid and consistent decisions and the removal of physical limits imposed by the human body may provide a decisive advantage. As Michael Byrnes contends in a highly topical paper, “a tactically-autonomous, machine-piloted aircraft […] will bring new and unmatched lethality to air-to-air combat.” Brought to its logical conclusion, he argues, “a single [such aircraft] with a few hundred rounds of ammunition and sufficient fuel is enough to wipe out an entire fleet” of manned aircraft.

If such visions should prove even partially accurate, the most advanced air forces could find themselves outmoded by an autonomous air-to-air system, much in the way that the Dreadnought relegated battleships that had been launched only years before to virtual obsolescence. While the mechanics would be quite different, AWS may also provide radical advantages in overcoming enemy defenses and executing air-to-ground, anti-ship and other missions in nonpermissive environments, with similar consequences. The resulting arms dynamics could have both qualitative and quantitative elements, even though the former are likely to dominate initially. They may pit offensive systems against other offensive systems, offensive against defensive, or both.

Modernization would probably occur all around, and might be only moderately competitive once a baseline autonomous capability has been established. More consequential would be the destabilizing effect upon individual military balances, in which AWS capabilities might be unevenly distributed to begin with and several rounds of catch-up and advantage-seeking are required to reestablish a stable balance at a higher level of capability. This would especially be true of constellations in which both sides are motivated by an underlying geostrategic rivalry, and have considerable economic and military potentials that allow them to sustain the competition over time. While this could include other great- and medium-power constellations as well, it may apply to the United States and China, in particular.

Of course, there are many scenarios for the adoption of AWS, the most likely of which involves a gradual slide towards autonomy, in which manned, unmanned, semi-autonomous and eventually fully-autonomous systems coexist for many decades, and sudden obsolescence does not occur. What effects a more leisurely modernization process would imply are difficult to fathom, but the potential for instability would likely remain.

The Stimulus to Strike

Irrespective of the pace at which the adoption of autonomous weapon systems would proceed, the introduction of such systems—particularly those with offensive missions—could have a significant impact on actors’ conduct in a crisis. While crisis behavior depends on a host of contextual factors, it is widely accepted that force structure (and posture) plays an important role in stabilizing, or further destabilizing, interstate relations at the brink of war. As a recent study has once again brought to the fore, forces that effectively support crisis diplomacy should be potent without being excessively vulnerable, and should pose a credible threat without instilling fears of a crippling “bolt from the blue.”

Could autonomous weapons contribute to the pressure to use force in a crisis in ways that manned or remotely controlled systems do not? At first glance, it would seem that while they are merely kept on alert in their bases, AWS do not provide any first-strike incentives that a human-controlled system of similar configuration and capability would not also provide. However, considering that autonomous weapons may be endowed with capabilities that are far beyond those of human-controlled systems in some areas, an actor may see a window of opportunity to disable the systems before they begin to act autonomously, and therefore become much more difficult to find and defeat. Knowing that its systems are vulnerable before they are activated, the other party may feel pressured to activate them early, which might induce the first actor to activate his systems as well, and so on. Another set of incentives to “jump the gun” would arise if autonomous systems possess any specific vulnerabilities that provide a premium on early kinetic or nonkinetic attack; e.g. if they are serviced by a few exposed pre-mission programming centers. However, such vulnerabilities are not inherent in the concept of AWS, and they are avoidable in principle.

Introduction