Robotics, air power, and “just” war

The rapid evolution of unmanned combat aerial vehicles calls these principles into question. An open letter signed by Stephen Hawking and Elon Musk recently warned against the development of autonomous military robots. Hawking and Musk asserted that artificial intelligence has developed to the point where the potential for UCAVs to select and engage targets without human input is “feasible within years, not decades”.

In 2011, the UK Department of Defence issued a report stating there was a consensus that artificial intelligence in unmanned aircraft would “be achieved … [in] more than five years and less than 15”.

Evidence of progress includes prototype aircraft that can autonomously fly long-distance missions (BAE Systems’ Taranis) and takeoff and land on aircraft carriers (Northrop Grumman’s X-47B).

The imminent rise of robotic aircraft raises a range of ethical issues outside the scope of the prevailing law and norms of military conflict.

Firstly, removing the risk to our own pilots from decision-makers’ calculations could make nations more likely to go to war. In 1862, after the Battle of Fredericksburg, Confederate General Robert E Lee remarked: “It is well that war is so terrible – otherwise we would grow too fond of it”.

Similarly, the great Prussian strategist Carl von Clausewitz suggested that the risk to one’s own forces is a significant factor in limiting the aggressive policies that contribute to war’s brutality. Even for the victor, warfare historically has involved loss, terror and suffering. It has rarely been viewed as an unambiguous good, increasing the likelihood that it will be employed only as a last resort – as prescribed by jus ad bellum.

Although this theory has yet to be fully tested, it is supported by early indications. In 2011, the White House argued that Congress did not need to authorise US drone operations over Libya after the initial 60 days approved by President Obama because the conflict did not “involve sustained fighting or active exchanges of fire with hostile forces, nor … the presence of US ground troops, US casualties, or a significant threat thereof”.

This suggests governments are likely to apply lower standards to wage war when there is no risk to their own troops, potentially weakening their compliance with the principles of proportionality and military necessity.

Robotic aircraft distance people from conflict.

A further issue associated with robotic aircraft is that they distance people from conflict. Human recognition of the suffering caused by war is likely to diminish accordingly.

When fully autonomous aircraft are operational, the moral consent of pilots to engage in war will no longer be required. Currently, pilots who believe a war to be unjust can refuse to fly. In 1972 the refusal by American B-52 pilot Captain Michael Heck to fly bombing missions over North Vietnam generated public debate over the moral issues. Missions flown by artificial intelligence will no longer be subject to a pilot’s scrutiny.

It is uncertain whether aircraft controlled by artificial intelligence will be capable of distinguishing between combatants and non-combatants. This concern is exacerbated by air power’s increasing prominence in counter-insurgency operations, where combatants and civilians often occupy the same spaces and wear the same clothes.

Further complexity is added by applying the test of proportionality to the potential military gain of an action against the possible harm to civilians. The inability to incorporate factors such as empathy for civilians, or to apply a complex and accurate understanding of an action’s consequences, limits a computer program’s (robot’s) ability to resolve such tests. A robot cannot be expected to assess the value of killing a particular insurgent, let alone balance it against the potential harm to civilians or determine the reliability of the information on which the attack is based.

Finally, when robotic aircraft inevitably cause civilian casualties in violation of international human rights and humanitarian law, the lack of human control creates accountability problems. With nobody clearly at fault, there can be no retribution for victims and no deterrence of future crimes.

UCAVs represent an opportunity to eliminate casualties for one side. In many respects this is a welcome development. However, we need to consider updating the law and norms of armed conflict to guide air forces’ transition to robotic warfare in line with the theory of “just” war.