The UN special rapporteur for extrajudicial, summary, and arbitrary executions Cristof Heyns has recently published a report on the dangers of lethal autonomous robots (LARs) currently under development in several countries.1 In this report the UN rapporteur stated:
‘The experience of the two World Wars of the last century may provide insight into the rationale of requiring humans to internalize the costs of armed conflict, and thereby hold themselves and their societies accountable for these costs. After these wars, during which the devastation that could be caused by modern technology became apparent, those who had personally taken the central military decisions resolved, “in order to save succeeding generations from the scourge of war”, to establish the United Nations to pursue world peace and to found it on the principles of human rights. While armed conflict is by no means a thing of the past today, nearly 70 years have passed without a global war. The commitment to achieve such an objective can be understood as a consequence of the long-term and indeed inter-generational effects of insisting on human responsibility for killing decisions.’2
The report stressed that the current time might represent the best chance to implement regulation or a complete ban of LARs since this would be much more difficult after such weapons have been introduced. Heyns points out that there “is clearly a strong case for approaching the possible introduction of LARs with great caution.” What the report eventually suggests is a moratorium on the development of LARs on a national level so that the policy and legal issues arising from LARs in the long term can be better evaluated.3
I think Cristof Heyns proposal for a moratorium is well-intentioned and would make a lot of sense. Unfortunately, there are lots of problems in terms of what the exact objectives behind this proposal are, how LARs can be legally defined, and in terms of what can be realistically achieved by such a moratorium. The report of the special rapporteur oscillates between different rationales for a moratorium and a possible ban of LARs. This creates confusion with respect to what the main objective behind a moratorium would be and what the real problems are with respect to robotic/automated weapons systems. On the one hand, the report outlines the issues that deal with the technological push towards autonomy and the technological limitations that raise doubts about the ability of such systems to comply with the demands of international law. On the other hand, the report also heavily criticizes the current use of teleoperated drones for targeted killings around the world and the prospect of robots and drones being used for political repression by ruthless elites. These are all very reasonable arguments that I have made myself, but if one pursues an international arms control agreement or even voluntary national moratoria on the development of the technology, there should be some clarity behind the main objective. Is it about reversing the trend of weapons automation, is it about stopping the use of drones for targeted killings around the world, or is it about preventing the abuse of robotic technology for political repression in a domestic policing context rather than a context of armed conflict?
By including everything the special rapporteur has weakened his case, especially as he has not defined what exactly states should refrain from developing. As pointed out, there are already some highly automated weapons systems in use. States will not reverse decades of military technology development to go back fighting with bows and arrows. The term LARs does not seem to include armed drones as currently used by the US, Britain, and Israel. No moratorium on the development of LARs could possibly affect the practice of targeted killings, especially since the main idea behind targeted killing is not a robot deciding who to kill, but rather the idea of the robot as an executioner going through ‘kill lists’ that have been put together by the intelligence services. Sure, this process of compiling targeting lists seems to become increasingly automated as well, if one considers the ‘disposition matrix’ – a database that would be designed to manage and coordinate US efforts to kill or capture certain individuals.4 However, a ban of LARs would not address software from adding names to targeting lists based on the communication patterns or other activities that flag them to computers as potential terrorists. In any case, not much automation is needed for eliminating some rag-tag militia members and terrorists with no air defense systems.
Finally, the idea of a moratorium on LARs addressing the issue of reversing the trend of turning modern societies into high-tech police states, as it is happening in the US and elsewhere is even more hopeless. Police and internal security forces already have access to drones and the police use of drones for domestic surveillance and crime fighting will continue with or without an international agreement. Ironically, the types of weapons that are banned by international law and cannot be used in international conflicts, such as expanding bullets, can be perfectly legal to use in a domestic context. In addition, police drones would be equipped with so-called non-lethal or less-than lethal weapons like Taser rounds, so they would by definition not fall under the term LAR.
In terms of what can be realistically expected from a moratorium I am also rather pessimistic. The states that are leading in terms of technology would have the most to lose if the support a moratorium on the development of LARs because it would just mean that other countries would have greater opportunity to catch up technologically and nullify all the advantages currently held by a handful of states. Secondly, the most capable weapons systems are these days developed and sometimes even used in complete secrecy. The US military has ‘white’ R&D projects and then it has also ‘black’ R&D project that are potentially a lot more advanced and revolutionary of what we know of. For example, it has been revealed that the US is secretly working on very advanced unmanned stealth aircraft. Little is known about their capabilities or even purpose. Nations will try to evade and circumvent international arms control agreements by keeping their most advanced toys secret.
The idea of a moratorium of LARs looks initially like a good one, but I very much doubt anything will come out of it. Governments will always do whatever they can do and get away with it, unless they are persuaded that there is just no benefit in certain practices. The only way of preventing further automation of war is by proving to governments that there is no benefit for them doing so. In addition, it does make sense to aim for arms control agreements that limit the permissible numbers of highly autonomous weapons systems and to put strict limitations on the capabilities of highly automated systems in terms of range, payload, and endurance. Permitting militaries to use a few very smart autonomous weapons with limited capability under clearly defined specific circumstances would most likely produce better humanitarian outcomes in war than freezing military technology at the current state where militaries can use an unlimited number of dumb automated weapons.
- Cristof Heyns, “Report of the Special Rapporteur for Extrajudicial, Summary or Arbitrary Executions, Cristof Heyns,” UN General Assembly (A/HRC/23/47).
- Para. 96.
- Para. 118.
- Greg Miller, ‘Plan for Hunting Terrorists Signals U.S. Intends Adding Names to Kill Lists’, Washington Post, 24 October 2012.