Autonomous Weapons

Military robotics technology is growing in intelligence, capability, and autonomy. The sensing technologies, which are central to robotic intelligence, are increasing in sophistication. State of the shelf sensors can acquire and process information in the visual, auditory, tactile, and gustatory dimensions. State of the art sensors can gather and process human intentions. Wide-area surveillance sensors like the Gorgon Stare can film an area with a four kilometer radius during day and night operations from twelve different visual angles. Acoustic sensors can detect, localize, and prepare countermeasures to unfriendly gunfire. Chemical and biological warfare multi-sensing technologies can provide information on carbon monoxide, hydrogen sulfide, volatile organic compounds, oxygen, lower explosive limit, carbon dioxide and air particulates including airborne dust, smoke, mist, haze and fumes. The U.S. Department of Homeland Security’s MALINTENT sensor program features a series of sensors that read body temperature, heart rate, and respiration, among other non-verbal cues, in order to detect signals of aggression or distrust. New technologies are emerging at a rapid pace.

Today, most such robotic systems relay the information acquired to human operators, who then decide a course of action, but, as with the UK Taranis stealth jet, there is a desire to develop autonomy to the extent that enables machines to automatically decide what constitutes a target and, subsequently, what line of action should be taken upon a target. The evolution of robotic intelligence, capability, and autonomy promises a future where machines can learn, derive rules, and execute orders with minimum, if any, human input. Such technology may be a godsend in removing warfighters from dull and dangerous jobs, and it also may be a beast of burden in its ethical, legal, and social implications.

Ethically, there must be balance between acting upon the basis of a duty to provide security and ensuring the relative rewards outweigh the risks. Some interpretations of international law sanction and encourage the employment of lethal robots, while other interpretations vehemently oppose machines capable of autonomous force. On the one hand, these machines can be trusted to make decisions that are not influenced by the stresses and strains of war. This could result in reduction of non-combatant casualties. On the other hand, it is questionable whether or not machines should be enabled to kill without authorization from a human operator. It is also questionable whether or not machines can embody the discriminatory intelligence necessary to guarantee, or minimize, civilian casualties.

The development of autonomous machines deeply impacts the identity, values, beliefs, and perceptions of societies. Also, the inverse is true in that the identity, values, beliefs, and perceptions of a given society influences whether or not and to what extent autonomy in lethal machines is seen as a legitimate, or even necessary, instrument of state power. For instance, although the ability for autonomous robotics to reliably discriminate between combatant and non-combatant is important for US policy makers, not all actors will employ this same instrumental rationality. Different cultures will judge the costs and benefits of the development of these systems differently. Rogue regimes or non-state actors may indiscriminately develop and utilize autonomous weapons systems.

This discusses the conceptual, technical, political, social, ethical, legal, and military dimensions of autonomous weapons. It discusses the state of the art, the state of the shelf, and a probable future wherein machines can learn to derive and execute their own rule sets. It recommends typical analyses that could be conducted for the purposes of current policy, as well as strategic assessments that could be conducted for the purposes of preventing strategic surprise. Although far from an exhaustive account of autonomous weapons, this paper makes a contribution to the available literature.