There are at least three major ethical issues associated with autonomous weapons systems: (1) Could and should machines be enabled to make sensitive decisions about human life? (2) Does the United States have an ethical imperative to develop such autonomous systems? (3) What do we do if the (moral) cost of such development outweighs prospective benefits?
At this point, machines can only follow the rubric of pre-programmed constraints. Relative to state of the shelf technology, whether or not machines should be enabled to judge who to kill and who not to kill on their own, without a man in the loop, may be a matter of how well policymakers, or whoever is approving the employment of such systems, trust in discriminatory intelligence of these machines. The discriminatory intelligence of these first generation autonomous machines is only as good as the code that the programmers establish as the basis for the functionality of the machines. For those interested in the principals of proportionality and discrimination, which are staples of Just War Doctrine, it is imperative to ensure that these machines can distinguish between combatants and non-combatants with a ‘high’ degree of accuracy before being let loose in theater. The benchmark metric for determining ‘how much discriminatory intelligence is enough’ has yet to be defined in any official capacity. As an advocate of Just War Doctrine, and as a pillar of customary international law, the United States has an ethical obligation to maximally comply with the principles of discrimination and proportionality. Warfighters must discriminate between combatants and non-combatants. Loss of civilian life must be minimized. There is an implicit responsibility to employ whatever means maximize compliance with these principals; whether that means appropriately educated human beings or autonomous machines.
Implicitly, there is also a responsibility to minimize the adverse affects of war on warfighters. By adverse affects is meant the full gamut; from stress and strain to wounds and death. So long as they can satisfy the principals of discrimination and proportionality, if machines can save lives then there is an ethical obligation for machines to be so employed. To an extent, robots have been employed in war for the purposes of saving human life, such as the PackBot from iRobot, which goes on missions with the objective of locating and detonating Improvised Explosive Devices. In this capacity, robots have saved countless lives. Lives could be saved if robots were discriminate enough to help soldiers in gunfights.
From a big picture perspective, in terms of ensuring minimum casualties, there is an imperative to end wars as soon as possible. Therefore, wartime policymakers have a responsibility to assess whether or not, and to what extent, robotics can facilitate victory. From the military planning side, it is necessary to incorporate such technologies in a way that helps, rather than hinders, operational effectiveness.
Also from a big picture perspective, it is important to consider whether or not and to what extent the development and incorporation of robotics into war affects the sheer quantity of war. For instance, if state or non-state actors can cost-effectively utilize semi-autonomous or autonomous robotics for wartime operations then wars may become more plentiful than before. This train of thought would stick for countries, especially developed countries, which would go to war more, but who are dissuaded from doing so at least in part due to adverse public reaction as regards risking human lives. Further, robotics could be used by enemies of a state, or disenfranchised groups, to stage and carryout remote terrorist attacks, a type of terrorism that would proceed without the element of human sacrifice. These are some of the risks associated with autonomous robotics technology. The United States, among others, has a moral responsibility to assess and militate against the potential threats posed by such technology.
