NATO limiting leadership and active implementation of AI mechanisms now --- has shifted development to private industries and adoption to individual countries
Burt ‘21
(Peter, “NATO's new AI strategy: lacking in substance and lacking in leadership,” pg online @ //um-ef)
NATO's ethical principles for the responsible use of AI, though welcome, raise several issues. Such statements of principle are now commonplace in the corporate sector and are increasingly being adopted by governments on both a unilateral and multilateral basis. NATO's principles are similar to principles adopted by the US Department of Defense for the ethical use of AI: indeed, in some places the wording is the same as that in the Department of Defense principles. As with many such statements of principle, the NATO principles have no coherent means of implementation or enforcement. Their successful adoption will in many ways depend upon leadership and political and military culture, which is different in each of NATO's 30 member states. Will Turkey, for example, which has been a keen proponent of automated warfare and by some accounts has already developed AI-based loitering munitions with an autonomous capability to identify targets, be willing to follow the same rules as the USA? And would the USA under a second Trump administration follow the same rules as a Biden administration? In the absence of any binding enforcement mechanism NATO's principles may provide useful for public relations purposes but are likely to be less useful in preventing harm to humans, particularly those in the Global South who are already in situations of conflict, and historically marginalised groups. Despite the plethora of corporate statements on ethical principles, those working in the tech sector are sceptical about the prospect that ethical AI design will be adopted as a norm over the next decade. A non-random poll conducted by the Pew Research Centre in early 2021 found that 68% of experts in the field thought that ethical principles focused primarily on the public good will not be employed in most AI systems by 2030. Their concerns recognised that the main developers and deployers of AI are focused on profit-seeking and social control, and that global competition will matter more to the development of AI than any ethical issues. This latter factor will very much influence NATO's future adoption of AI. Although NATO's ethical AI principles are stated to have been developed on the basis of “Allied approaches and relevant work in applicable international fora”, it is not clear whether they draw on views from the wide range of professional disciplines necessary to develop a representative and rounded view of the ethical pitfalls and risks associated for AI, or from a diversity of perspectives. There has certainly been no open consultation on their formulation, 4 and it is possible that the principles may only represent the perceptions of a relatively narrow range of predominantly white male technocrats and military planners drawn from within NATO, member governments (notably the USA), and the arms industry. Ordinary people, particularly marginalised groups and those in the Global South, will ultimately face the consequences and impacts of NATO's decisions on AI systems, yet the public have certainly not been involved in making decisions on this strategy, which will set the framework for NATO's future AI choices. NATO's AI strategy does not discuss the development of AI-driven autonomous weapon systems – a significant omission, given the ethical issues that this application of AI would raise and the challenges to human rights that such weapons would pose. As a bare minimum, the strategy could—and should—have endorsed the 'Guiding Principles' on emerging technologies in the area of lethal autonomous weapons which have recently been adopted by consensus by High Contracting Parties to the Convention on Certain Conventional Weapons (CCW). Despite increasing pressure for a ban on the development and use of lethal autonomous weapons, key NATO states including the USA and UK have been lukewarm about the need for a ban, arguing that current laws of war are adequate to regulate any such weapons. This seems optimistic, given the rapid development of AI technology and the push by some states over recent years to redefine the rules that govern use of armed force to suit their own purposes. To date no NATO member states have supported the call for a ban on lethal autonomous weapons. If NATO members were serious about ensuring that the military use of AI adheres to international humanitarian law and human rights law, they would call for and engage in negotiations on a legally binding instrument on autonomous weapons systems at the CCW; which is expected to take a historic decision on this issue at its Review Conference in December 2021. Investment and innovation in artificial intelligence is being led by the private sector, and not by the world’s militaries. Recognising this, NATO's AI strategy places a premium on engaging with “start-ups, innovative small and medium enterprises, and academic researchers that either have not considered working on defence and security solutions, or simply find the adoption pathways too slow or restrictive for their business models”. NATO wants to “make defence and security a more attractive sector for civilian innovators to partner with”. Aiming to capture the civilian tech sector in this way risks increasing the influence of the military-industrialpolitical complex to an even greater extent than is already the case. Will this really help advance NATO's liberal and democratic values? Conclusion Although NATO's AI strategy contains a few worthwhile nuggets, on balance it is tame and unambitious. It is depressingly clear that NATO sees AI as basically another way of using technology to wage war more effectively, and is not willing to show any real leadership to mitigate the risks that AI poses to human rights and dignity. War is, after all, the highest area of risk when it comes to the potential for human rights abuses, yet NATO's strategy says nothing about measures to effectively govern military AI and autonomous weapon systems. AI has the potential to help humanity tackle intractable 'wicked problems' such as climate change and unsustainable development, and tackle the course of dangerous behaviour which is threatening the survival of our species. But this will only happen if it is employed under wise and decisive human leadership, otherwise it may dramatically compound the problems we face, threatening international security and human rights even more. NATO, unfortunately, seems unwilling to provide that leadership. Acknowledgement: The author would like to thank Wanda Muñoz for advice and comments on an earlier draft of this paper. Author: Peter Burt is a researcher at Drone Wars UK working on issues relating to artificial intelligence and autonomy and their role in the future development of drones. Before moving to Drone Wars UK Peter was Director for the Nuclear Information Service for six years, and he is also a Trustee the Nuclear Education Trust.