Militarizing Artificial Intelligence

Theory, Technology, and Regulation

Nik Hynek, Anzhelika Solovyeva

• Publisher: Routledge
• Year: 2022
• ISBN:
  9781000609301
• Edition: 1
• Topic: Politique et relations internationales
• Subtopic: Sécurité nationale

This book examines the military characteristics and potential of Artificial Intelligence (AI) in the new global revolution in military affairs.

Offering an original perspective on the utilization, imagination, and politics of AI in the context of military development and weapons regulation, the work provides a comprehensive response to the question of how we might reflect on the AI revolution in warfare and what can be said about the ways in which this has been handled. In the first part of the book, AI is accommodated, both theoretically and empirically, in the strategic context of the 'Revolution in Military Affairs' (RMA). The book offers a novel understanding of autonomous weapons as multi-layered composite systems, pointing to a complex, non-linear interplay between evolutionary and revolutionary dynamics. In the second section, the book provides an impartial analysis of the related politics and operations of power, whereby increases in military budgets and R&D of the great powers are met and countered by advocacy networks and scientists campaigning for a ban on lethal autonomous weapons. As such, it moves beyond popular caricatures of 'killer robots' and points out some of the problems which result from over-reliance on such imagery.

This book will be of much interest to students of strategic studies, critical security studies, arms control and disarmament, science and technology studies and general International Relations.

Part I Artificial Intelligence and Dynamics of Military Transformation

1 Artificial Intelligence and the Revolution in Military Affairs

DOI: 10.4324/9781003045489-3

Introduction

We are told that we stand on the cusp of another revolution in military affairs (RMA). It is being brought about by a rapidly evolving, transformative, and potentially disruptive technology commonly known as artificial intelligence (AI). Dreams of human-like machines and the earliest documented examples of automata date back far into the past, but the first real efforts to build intelligent machines originated in the mid-twentieth century (Nilsson 2010: 71; Cave and Dihal 2018). It is often ignored, however, that there is no such a thing as AI in a complete form. This is, in fact, an umbrella term for a broad range of advanced technologies which allow the mimicry of human abilities in machines and breathe life into what we know as AI. Such technologies include but are not limited to machine learning techniques, artificial neural networks, computer vision, multisensor data fusion, as well as algorithms for natural language processing, pattern and speech recognition, knowledge representation, and reasoning (Nilsson 2010). The Internet of Things and Big Data facilitate progress towards AI (Hallaq et al. 2017). AI technologies underlie, among other things, the intelligentization of weapons, battlefield networks, intelligence, surveillance and reconnaissance capabilities, military logistics, and war games. This is where our focus lies and what this book analyses under the rubric of the AI-RMA. Since the concept of RMA is often evoked in relation to AI, this book begins with an analysis of whether and how AI fits in with what is conceptualized as RMA. In particular, we seek to understand if the existing intellectual framework can be applied to reflect on the transformative potential of AI. Another goal emerges as the analysis proceeds: to bring the existing theorization of RMA closer to reality.

RMA analytical framework: Accommodating AI

This section reviews the existing literature and unravels the phenomenon of RMA. The focus is on the readiness of its conceptual and historical dimensions to accommodate and explain the revolutionary potential of AI. It is fair to begin this discussion by noting that there has been no consensus on the definition of RMA. However, and here we part with Shimko (2010: 14), differences are often a matter of degree. Though making the application of the concept itself less straightforward, they do not undermine its analytical significance and explanatory potential. What follows is a systematic inquiry into the various existing theories of RMA. In particular, we scrutinize and report on progress towards an integrated analytical framework. The following aspects are considered: the role of technology in generating military change, the nature of revolutionary outcomes, and the respective military history.

Components of change

It is broadly accepted in the existing literature that technology plays an important role in facilitating military revolutions. Yet all parties concur that it is not the only driving force (van Creveld 1991: 32; Toffler and Toffler 1993: 31–34; Rogers 2000: 30; Murray and Knox 2001: 12; Cohen 2004: 399; Adamsky 2010: 1). However, theorists differ in views as to the degree of technological impact. Some scholars treated technology as the necessary condition (Krepinevich 1994). They argued that it sets the parameters of the possible (Boot 2006: 43). Others were less resolute in their views. They assumed that technology usually (e.g., Fitzsimonds and van Tol 1994: 25) normally (e.g., Morgan 2000: 134) or often (e.g., Sloan 2002: 25) drives change in military art. Still more sceptical writers note it only sometimes and not always does so (Hundley 1999: 14; Horowitz 2010: 22). In other works of this kind, technology was conceived of as a relatively insignificant factor (Murray 1997). They believed it rarely plays a decisive role (Murray and Knox 2001: 180). Two approaches stand out as the leading sources of theorization. They differ with respect to the role of technology and assign different weights to different factors in the overall composition of change.

One body of literature takes technology seriously and theorizes the requisites of change. In spite of differences in formulation, their proposed set of indicators basically includes technological change, operational innovation, and organizational adaptation (Fitzsimonds and van Tol 1994: 25–29; Krepinevich 1994; Hundley 1999: 14–22 and 33; Morgan 2000: 135–138; Gilli and Gilli 2016: 56–60). It is well-tailored to analyse technology-driven revolutions, including both military-technical revolutions (MTRs) and revolutions in military affairs. The former dates back to the writings of Soviet military thinkers in the 1970s and 1980s (Krepinevich 2002: 1; Thompson 2011: 85); the latter is a more recent term and differs in the increasing recognition of doctrinal and organizational factors over the primacy of technology (Fitzsimonds and van Tol 1994: 25–26; Hundley 1999: 14–15; Morgan 2000: 134–135; Sloan 2002: 24–25). Adamsky (2010: 5) went even further to argue that there is also strategic culture, which provides the decisive context for one’s RMAs. The logic of MTRs, though shelved for the present, sometimes manifests itself in different and more neutral terms such as military revolutions (Krepinevich 1994) or military innovations (Gilli and Gilli 2016: 82).

The other body of literature theorizes potential sets of indicators to get closer to reality. For them, technology is only marginally significant. Although it helps them to better capture the variety of military revolutions, there has been no agreement on the nature and composition of factors in the equation. For instance, Rogers (2000: 22–24) analysed revolutions in military affairs as changes in how war is fought, including technical, tactical, and strategic innovations. However, he argued that such changes in certain, but by no means all, cases have extremely wide-ranging social, economic, and political implications. Whenever it is the case, he used another term: (full) military revolutions. Knox and Murray (2001: 6–12) focused on (great) military revolutions recasting the entirety of society, the state, and the system of military organization. However, they acknowledged the existence of less profound transformations that either accompanied or followed such revolutions. They studied them as revolutions in military affairs and as a complex mix of tactical, organizational, doctrinal, and technological innovations. These two approaches are similar in terms of viewing military revolutions as much broader structural changes compared to RMAs, but they apparently differ in how they theorize the relationship between these two terms and phenomena. There are many more approaches in fact. For example, Toffler and Toffler (1993: 31–34) inquired into the transformation of society under the rubrics of civilizational change. Such large-scale transformations, in their opinion, entail (true) military revolutions. The latter involve profound changes in armed forces at various levels from technology, doctrine, and organization to tactics, strategy, logistics, and culture. Their approach did not override the possibility of sub-revolutions, potentially limited to certain contexts but also part of the total ‘game’. Boot (2006: 26–28) drew attention to paradigm shifts associated with transformational technologies. He presented them as ages. Only when opportunities for new competitive advantages are exploited by societies and their armies do we experience military revolutions. The author associated them, inter alia, with changes in organization, strategies, tactics, leadership, training, and morale. Gray (2006: 16–28) distinguished six different ‘contexts’ shaping the prospects of occurrence and the character of what he called revolutionary change in warfare. These are strategic considerations, socio-cultural trends, economic conditions, technological capabilities, politics, and geography. Horowitz (2010: 23–42) analysed military innovations and conceptualized them as the product of one’s choices and possibilities. Strategic choices, in his view, arise from geostrategic dynamics, international norms, domestic politics, and cultural openness. Strategic possibilities are, according to his interpretation, constituted by one’s capacities in terms of technology, finance, and organizational capital.

Two overarching approaches are presented but, as we show, individual contributions also considerably differ from one another. Further confusion apparently comes from the mismatch in their terminologies. How does this knowledge help us understand the transformative potential of AI? Since the consensus is that technology itself is not enough to trigger a revolution, we are instructed to search for other components of change. This is a serious challenge, however, because the existing literature raises more questions than it answers. There is no coherent, integrated guidance on how military revolutions originate, what they entail, and how their driving forces interplay. Chapter 2 will bring to light further shortcomings. It will focus on the existing conceptualizations of revolutionary technologies and reveal their limits in light of AI. It is for this reason that we intentionally keep from discussing it here.

Revolutionary outcomes

The assumption that military revolutions are paradigm shifts in the character or conduct of war runs throughout the literature (Krepinevich 1994; Hundley 1999: 9; Gray 2002: 67–68; Sloan 2002: 3; Horowitz 2010: 22–23; Shimko 2010: 9–11). It is accompanied by another assumption that we know as common sense, that the nature of war as organized violence for political ends, as well as its inherent qualities such as friction, fog, ambiguity, chance, uncertainty, and confusion do not change (Gray 1996: 8; Murray 1997; Murray and Knox 2001: 178–179; Sloan 2002: 30; Shimko 2010: 9; Hoffman 2017/2018: 31). However, there is no agreement on how to detect that a complete revolution has taken place. Cohen (1996: 43–51) operationalized it as change in forms of combat, including the fundamental relationship between offence and defence, space and time, fire and manoeuvre, as well as in the system of military organization, the nature of command, and the relative balance of power. Elsewhere, he (2004: 403) recommended to consider it a revolution if military forces, battle processes, and outcomes look fundamentally different. Rogers (2000: 22) assumed that a revolution manifests itself as ‘the ease with which “participating” armed forces can defeat “non-participating” ones’. According to van Creveld (1991: 14–15), it is change in the causes and goals of war, blows with which campaigns open and victories with which they end, modes of organization, command and leadership, methods of planning, preparation, execution and evaluation, strategies and tactics, operations and missions, capabilities, conceptual frameworks employed to think about war, and relationships between armed forces and societies. Generalization is barely possible. Since failed or incomplete military revolutions are rarely studied, we also face the problem of false positives (Hundley 1...