Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

Starting with logistics pilot projects spills over to effective NATO AI cooperation more broadly---it creates buy-in and prepares the Alliance technologically, socially, and politically

Andrea Gilli et al 20, Senior Researcher at the NATO Defense College where he works on issues related to technological change and military innovation. He has been visiting and postdoctoral fellow at Johns Hopkins University and Columbia University as well as Stanford University (where he remains an Affiliate) and Harvard University. Mauro Gilli is a Senior Researcher in Military Technology and International Security at the Center for Security Studies (CSS) at the Swiss Federal Institute of Technology, ETHZurich. Before joining CSS, he was a post-doctoral fellow at the Dickey Center, Dartmouth College. Ann-Sophie Leonard is a former Mercator Fellow on International Affairs, focusing on the intersection of international security and technology. She was a Visiting Fellow at the Research Division of the NATO Defense College in Rome from September to December 2019. Zoe Stanley-Lockman is an Associate Research Fellow in the Military Transformations Programme at the Institute of Defence and Strategic Studies at the S. Rajaratnam School of International Studies in Singapore. “NATO-Mation”: Strategies for Leading in the Age of Artificial Intelligence. NDC Research Paper No.15 – December 2020 //pipk

NATO and its Allies have strong reasons to adopt a pragmatic approach, which can be conveniently defined as 3S: small + secure = success. The rationale is as follows: • Small: small projects are unlikely to generate internal opposition or face major challenges; • Secure: secure projects are also at lower risks of technical or organizational failure; • Success: success is thus more readily within reach.

According to Google Brain’s co-founder Andrew Ng, for instance, it “is important for [the] first few AI project to succeed rather than be the most valuable AI projects”. This means they should be “technically feasible” and have “a clearly defined and measurable objective that creates business value”.134 Greg Allen, now at the US Department of Defense’s Joint Artificial Intelligence Center, reiterates this message from a different angle. He considers that many “AI projects fail not because of the technology, but because of a failure to properly set expectations, integrate with legacy systems, and train operational personnel”.135 Militaries have learned these lesson in previous instances of technological change: historically, many breakthrough technologies generated opposition, they did not deliver on their promises, and the increases in costs and time requirements further undermined their acceptance.136 As such, a pragmatic approach to small and feasible projects minimizes the chance that political or technical failure might slow down the Alliance’s adoption of AI.

A pragmatic approach to “NATO-mation” follows two key assumptions. First, many discussions about AI currently focus on the tension between narrow versus general AI. Some believe that the age of “the singularity” – where general AI will be superior to human intelligence – is near. Others believe that this is a remote dream and that, for the foreseeable future, intelligent machines will be able only to conduct limited, specific tasks.137 For the moment, the Alliance should just look at “narrow” AI. Second, the approach pursued by the Alliance should encompass a wide range of uses, and go beyond discussion limited to lethal autonomous weapons systems. The debate on lethal autonomous weapons systems that has taken place over the past few years has in fact distracted attention from the military relevance of AI and its broader implications for the international system. As David Judge, Vice President of SAP Leonardo, puts it, the conversation built around the question “Are we going to have red-eyed killer robots?” is not useful.138 Indeed, even insofar as the lethal debates are important, these are largely regulatory questions decided at the national level, rather than at NATO level. NATO would therefore be better advised to pursue the lines advanced by Judge: we “should instead focus on pragmatic uses of AI that are already here and ready”.139

While intellectual debates about these two assumptions are relevant to long-term patterns of adoption, current adoption efforts should focus on exploiting narrow AI below the tip of the spear. Such exploitation depends on investments in complementary assets, goods, services, and changes to business models. The San Francisco-based consultancy Altimeter contextualizes these investments in its four main phases of AI adoption: exploration, experimentation, formalization and integration, depicted in the diagram below.140

When it comes to defence and AI, most NATO Allies sit between exploration (phase 1) and experimentation (phase 2). Only a few have eyed formalization (phase 3), and even fewer have started looking at integration (phase 4). This last phase coincides with what in business is generally known as enterprise adoption: when a technology becomes pervasive in the work of an organization. It is important to highlight, however, that at this stage, in the commercial world, even the most AI-intensive, ML-savvy and data-hungry companies have hardly made it properly to phase 3.141 Microsoft, Google, Amazon, Baidu and Facebook can deploy advanced AI solutions because they have access to an unparalleled amount of data and top talent, and because they have invested billions and years in AI. Other actors that lack these resources and capabilities should initially be more focused.142

Security analysts and scholars are primarily concerned that first movers in the AI race could reap the benefits of this technological transformation, thus forcing others to a condition of technological dependence and military vulnerability. Business scholars advise caution, as the conditions for first mover’s and second mover’s advantage to take place are quite specific.143 What matters for this discussion is that NATO and its Allies neither have to rush into AI nor have the luxury to wait: they need to prepare, technologically but also socially and politically, in order to avoid potential backlashes or resistance. In order to pursue this goal, one strategy consists in adopting a pragmatic approach based on a set of small and secure pilot-projects that, as such, are more likely to achieve success. Three potential areas are:

Logistics. General Omar Bradley famously said that amateurs study tactics, professionals study logistics. Logistics is, in fact, a critical element of military operations – although often neglected among the public, and even among some experts.144 Logistic capabilities can in fact give significant combat advantages.145 This is a familiar story to NATO, whose forces have, at times, enjoyed massive logistic advantages vis-à-vis some of their adversaries.146 Logistics can be divided into layers: strategic, operational, in-theatre and on the battlefield. There is definitely room for employing AI at every level. Launching AI pilot projects in last-mile logistics (e.g., on the front line), however, has significant benefits. In particular, if Amazon or Domino Pizza aim to rely on automated drone-based delivery systems, there is no reason why NATO Allies cannot at least try a similar approach when it last-mile resupplies.147 The advantage of last-mile logistics is that, lying at the end of the tail, it does not entail major infrastructure investments. However, it is complex, difficult and often also dangerous. For this reason, even small improvements in this realm are going to have significant effects. NATO and its Allies could fund a project, or launch a DARPA-like Grand Challenge, to develop an autonomous system for last-mile logistics. DARPA’s Grand Challenges are open calls to the public at large for participating in a competition with a prototype. In Europe, the Joint European Disruptive Initiative has tried to replicate on a smaller scale DARPA’s model, with remarkable success so far.

Sign in to recut, update, or save this card.