Reading Security Imaginaries as Fantasies – Loss,Desire,and Enjoyment in the Military Quest for Explainable AI

Abstract

What does the Lacanian notion of fantasy offer to the study of security imaginaries? The article answers this question by introducing a fantasmatic reading strategy illustrated by a case study of the US narrative of ‘technological revolutions of war’ that has recently been fueled by a growing demand for ethical and explainable artificial intelligence in military applications and weapon systems. The article offers a Lacanian comment to the expanding International Relations literature on security imaginaries. It demonstrates how a fantasmatic reading encompasses both a discourse analytical tracing of background understandings employed by many security imaginary scholars and an affective tracing at the margins of discourse that captures the force with which subjects continue to invest in – and patch the constitutive gaps of – a security imaginary. In studying security imaginaries through fantasies, we propose zooming in on three analytical moves: analyzing the continuing construction of a lost utopia in security discourses; following the specific objects of desire that is organized around its own inevitable failure; and locating the mode of enjoyment encountered at the boundaries of the socially acceptable norms.

Concevoir les imaginaires de la sécurité comme des fantasmes : La perte, le désir et la jouissance dans la quête militaire d’une IA explicable

Introduction

The belief that technological superiority leads to victory on the battlefield is ingrained in Western military thinking. ¹ The unwavering persistence of this belief, along with the institutionalized narratives that reinforce it, has prompted several International Relations (IR) and security studies scholars to explore the concept of a security or socio-technical imaginary, ² constituting a particular way of Western warfare. Sean Lawson refers to this constitution as the Western military imaginary ³ and Antoine Bousquet has dubbed it ‘The Scientific Way of Warfare’. ⁴

First introduced to the discipline by Jutta Weldes in the late 1990s, a security imaginary was defined as the repository of collective memory and lessons of the past – the cultural raw material that is mobilized when subjects come to grips with foreign and security politics. ⁵ In other words, imaginaries are generally understood – primarily through the classical works of Cornelius Castoriadis, ⁶ Benedict Anderson, ⁷ and Charles Taylor ⁸ – as background understandings informed by scripts, narratives, myths and representations of the self (and other), which shape and are expressed by social practices. ⁹ The concept has since been successfully applied to show how energy security practices, ¹⁰ populist and right wing practices, ¹¹ Western warfare, ¹² and European security practices ¹³ have come into being. However, among security imaginary scholars in IR it remains undertheorized why particular imaginaries are more appealing than others and why subjects continue to adhere to them.

Discourse analysts drawing on Lacanian psychoanalysis have already offered theoretical substance to explain why specific practices or regimes ‘grip’ the subject despite their inherent contingencies, and they have done so particularly through the notion of ‘fantasy’. ¹⁴ Fantasy, according to political theorists Jason Glynos and David Howarth, functions to give socio-political practices direction and affective energy. ¹⁵ The concept is gaining popularity in IR, mainly through attempts that explore the conceptual trading zones between Lacanian psychoanalysis and various IR sub-disciplines. ¹⁶ None, however, have explored the trading zone between the Lacanian fantasy and the emerging scholarship on security and socio-technical imaginaries. Additionally, only a few Lacanians have sought to develop a structured reading strategy that provides the non-Lacanian readers with an accessible step-by-step fantasmatic reading strategy for their respective IR sub-disciplines. The article addresses these gaps by exploring what is gained analytically from studying security imaginaries as fantasies.

By addressing this question, the article demonstrates, on the one hand, that a fantasmatic reading can encompass the discourse analytical tracing of scripts, narratives, and representations of self/other employed by numerous security imaginary scholars. On the other hand, it shows how to address the fact that security imaginaries are never fully available to discourse analysis and why the margins of discourse must be explored to grasp why a particular imaginary continues to grip the subject. Utilizing concepts such as lack, objet petit a, and jouissance, the article introduces a three-step analytical strategy. To capture the continuity and affective appeal of a security imaginary, we suggest, firstly, tracing the construction of a lost utopia in military and political security discourse, secondly, following the pursuit of specific objects of desire, and thirdly, locating the mode of enjoyment at the margins of socially accepted norms and conventions.

To exemplify the application of the fantasmatic reading strategy, the article introduces a case study examining the security imaginary within the US Defense, and specifically the narrative on technological revolutions of war. The case selection is deliberate as it serves to illustrate how fantasmatic readings can be operationalized and applied through a three-step analytical approach. Firstly, the analysis reveals that the US military pursuit of technological superiority constitute a lost utopia, successfully promoted from the initial technology offset strategy of the 1950s to the latest offset in 2014–8 and beyond. Secondly, the case study demonstrates that the military application of explainable artificial intelligence (XAI) functions as a prototypical object of desire – the perpetually elusive element – that sustains the fantasy of regaining the perceived lost US military superiority in the face of peer or near-peer threats. Thirdly, the study highlights excesses and irrationalities in speeches and statements by US Defense representatives, showcasing an affective realm at the boundaries of Law – a realm that facilitates initiatives and applications of military AI, operating outside legal restraint and oversight. Overall, the article contends that these affective dynamics are key to our understanding of how and why certain security imaginaries attain dominance, persist and are reproduced as meaningful narratives, despite their inherent contingencies and uncertainties.

The article starts out by engaging the existing literature on security imaginaries. Through a dialogue with the increasingly popular Lacanian-inspired IR and security studies, ¹⁷ we proceed to formulate a fantasmatic reading strategy that incorporates an ontology of lack – a fundamental inconsistency in the symbolic order and in the subject itself, resisting symbolization and fixation in discourse. ¹⁸ In this Lacanian reading strategy, the idea of the imaginary as the cultural repository of collective memories, myths, and narratives comes to the fore as the result of the endless, but ultimately impossible attempts to repair and reconcile the antagonisms and discrepancies of social identities as well as the formation of the ideological subject. Fantasy, then, is what allows the subject to perceive the imaginary as a meaningful and coherent order of things. Therefore, reading security imaginary as fantasy permits an investigation of the constitutive gaps and discontinuities within them, while shedding light on the subjects’ tireless effort to conceal such ruptures.

Affective Absences in the Security Imaginaries in IR

In her seminal work on the Cuban Missile Crisis, Jutta Weldes pointed IR away from a realist conception of national interest as an objective category and toward a social constructivist stance. ¹⁹ Drawing on Cornelius Castoriadis’ social imaginary, Weldes introduced the security imaginary as a ‘structure of well-established meanings and social relations out of which representations of the world of international relations are created’. ²⁰ Consequently, national interests are constructed by various institutions or a bundle of practices through an array of already available cultural and linguistic resources, termed the security imaginary. Such resources enable and constrain the representations that clarify who we are and who and what our enemies are as well as how to deal with them. ²¹ In other words, a security imaginary is the background structure of meaning within which foreign policy identity is discursively negotiated. ²² As a result, the analytical aim becomes to locate the signifying operations that constitute the cultural and linguistic resources, ultimately to explain how certain security actions and practices come into being.

In a similar vein, Joelien Pretorius, drawing on Charles Taylor’s social imaginary, ²³ explains the workings of military security imaginaries, where a perceived cultural match between societies and reinforcing security discourses ‘are important factors in the process by which one society’s security imaginary is infused by military models from other societies’. ²⁴ Both Sean Lawson and Dan Öberg echo the analytical endeavor to locate the key signifying operations underpinning a particular security imaginary. They point to war theories and the creative and planning-oriented representations of warfare, respectively, as those which substantiate, justify, and legitimate a particular Western military imaginary, and ultimately explain how global politics relate to the use of force. ²⁵ Similarly, Daniel McCarthy identifies the signifier technological innovation not merely as a means to an end for US national security, but as a shared background of meaning – the imaginary – upon which American national identity is reproduced. ²⁶ And Trine V. Berling, Izabela Surwillo, and Sandra Sørensen show how national narratives on nature and soil in Norway and independence (from Russia) in Ukraine are expressed in the two countries’ energy policies. ²⁷

The above examples illustrate well that certain discursive practices – often traced through various policy documents – are important for the constitution and reproduction of a security imaginary. Put differently, the discursive practices are signifying operations that explain the hegemonization of certain imaginaries. This is most explicitly explained by Raluca Csernatoni in her study on how EU institutions – through (floating) signifiers such as strategic autonomy and technological sovereignty – shape the public security imaginary. ²⁸ By merging Shiela Jasanoff’s idea of a socio-technical imaginary and Ernesto Laclau and Chantal Mouffe’s notion of a hegemonic discourse, Csernatoni introduces the hegemonic imaginary as a way of capturing the performative nature of security imaginaries and their political implications. ²⁹ Yet, what this move also highlights, is the fact that most scholars engaging with security imaginaries in IR are, in fact, treating these imaginaries as hegemonic discourses – as those meanings that have become (partially fixed) taken-for-granted through the articulations of contingent cultural and linguistic resources. While it is by no means wrong and certainly comes with benefits to move conceptually toward the imaginary, the move has simultaneously cloaked the limitations that come from treating security imaginaries as linguistic ‘background structures’ of shared and taken-for-granted meaning.

Laclau, the most influential voice promoting analyses of the production and implications of the taken-for-grantedness of hegemonic discourse, pointed to the need for an additional analytical step to grasp how structures become ‘common sense’ background meanings. He insists that analyses of discursive forms or structures neglect the force that accounts for the subject’s investment in these structures. ³⁰ To Laclau, the force that binds the subject is affect. While many discourse theories have come to embrace the intricate relationship between affect and discourse, ³¹ studies of security imaginaries have yet to follow suit. ³² The article makes up for this.

Laclau and many of those IR scholars who have turned to affect find conceptual inspiration for this move in Lacanian psychoanalysis. ³³ This is not to say that the turn to Lacanian psychoanalysis involves an ontological departure from the poststructuralists or critical constructivists in IR who analyze security imaginaries as discursively (re)produced taken-for-granted meanings underpinning foreign (and security) policy identities and action. Such a departure is not imperative. ³⁴ Rather, the affective move, through Lacan, adds to existing studies of the reproduction of security imaginaries – seeking to capture why signifying operations resonate with the subjects that enact and sustain particular security imaginaries. ³⁵ Yet, more radically, as this article proposes, Lacan’s conceptual edifice offers a useful analytical strategy for both reconstructing the form of a security imaginary and capturing the force with which subjects invest in it. We introduce such a strategy through the Lacanian concept – fantasy.

Several scholars have already introduced a Lacanian understanding of fantasy to various IR audiences. ³⁶ Here, most contributions successfully adapt their fantasmatic readings to the different IR subfields. Jakub Eberle, Andreja Zevnik, Marco A. Vieira as well as Catarina Kinnvall and Ted Svensson open up a conceptual trading zone between Lacanian fantasy and ontological security scholarship, particularly by proposing alternative ways of analyzing anxiety and the (postcolonial) Self. ³⁷ Moran M. Mandelbaum develops a fantasmatic framework to contribute to the IR discussions on nationalism, ³⁸ Henry Maher does a similar thing to add perspectives to debates on neoliberalism, ³⁹ and Jeppe T. Jacobsen does it to introduce the notion of transgression to surveillance studies. ⁴⁰ This article follows a similar path by developing a Lacanian reading strategy specifically for those interested in studying the workings of security imaginaries. Thus, the following introduces only those key concepts tied to the Lacanian notion of fantasy necessary for this particular analytical endeavor.

Fantasy – From Concepts to Reading Strategy

It is in the writings of Jacques Lacan that we find the most explicit and elaborate discussion of the imaginary as a specific register of subjectivity. It is not reducible to theoretical concepts such as dispositif and discourse or the psychological concept of the unconscious as a hidden or repressed reservoir of meaning, memories and emotions that can be revealed through an investigation and deciphering of the hidden layers of consciousness. ⁴¹ Yet, as will be elaborated shortly, the particular conceptualization of the Lacanian imaginary does not crystallize into ideas of coherent, self-enclosed imaginaries. ⁴² Rather, the Lacanian imaginary directs our analytical attention to a constitutive lack or absence of meaning at the heart of being a subject – an ontological split or gap – that is covered by the subject through its constant attempts to identify with the roles and positions offered by society.

One of the Lacanian strategies for studying how this constitutive lack urges the subject to constantly work – and inevitably fail – to identify (fully) with various roles and discursive positions in society is through the notion of fantasy. Hence, in a fantasmatic analysis we come to understand the ways in which the subject is engaged in a perpetual search of (self)identity, meaning and security, constantly desiring to fill the gap that separates it from itself. As noted by Solomon, ⁴³ the experience of lack ‘sparks an entire politics of desire and subjectivity. Thinking about the subject in this way entails not only lack itself, but more importantly how subjects attempt to cover over and negotiate this lack’. The Lacanian concept of fantasy is key here, precisely because it theorizes the built-in ambiguities and failures of such negotiations. ⁴⁴ Approaching security imaginaries through fantasy, however, requires a basic understanding of the theoretical foundation for Lacanian scholarship; that is, the tripartite register of the imaginary, the symbolic and the real.

Lack in the Tripartite Register

At its most fundamental, the Lacanian subject is always in search of itself. The search can be traced, first, to the register of the imaginary. In Lacanian psychoanalysis imaginary refers to that which allows the subject’s narrative of being an I. It is tied to the process through which one identifies with a (mirror) image of what seems to be a whole, complete, independent entity as itself. ⁴⁵ But it comes with an ambiguity, namely a ‘need to identify with something external, other, different in order to acquire the basis of self-united identity’. ⁴⁶ As a result, every self-image contains within itself an element of difference, a source of alienation, a lack. In other words, the imaginary always fails to provide us with a stable identity. Our only option is to search for one by becoming a speaking subject in the field of linguistic representation known as the symbolic. Hence, a Lacanian imaginary differs from the dominant use of ‘imaginary’ by security imaginary scholars, as it is not simply constituted by shared myths, narratives, and cultural tropes, but rather by a need to reconcile oneself with something else that is never quite within reach.

In Lacanian psychoanalysis, the subject’s search for itself is always mediated through the symbolic; that is, the socio-linguistic structures of society that condition the making of social and political identities. As such, our possibilities for grasping security, understanding it and formulating ways of dealing with it, are always structured by the symbolic. ⁴⁷ With the entry into the socio-linguistic order, the subject seeks a place which confirms its existence as a subject and does so by asking ‘what does the symbolic (Other) want from me?’ ⁴⁸ The symbolic responds with an interpellation – or hailing – of the subject as a citizen, consumer, soldier, etc., positions that are then assumed by the subject. ⁴⁹ Yet, alienation remains a basic condition also within the symbolic. ⁵⁰ The subject always feels as if it is more than the subject-positions provided, and the subject is, thus, constantly confronted with the unanswerable question as to why it is hailed into those particular symbolic roles, which are themselves contingent and arbitrary. ⁵¹

In many ways, the Lacanian symbolic shares similarities with security imaginaries in IR scholarship. It is at the level of the symbolic where scripts, narratives, myths and representations of Self and Other are formed. Yet, the Lacanian attention to the constant identification and alienation of the speaking subject offers additional theorizations on how contingent scripts, narratives, etc. are negotiated and reproduced. The subject’s discomfort with the inevitable alienation – the confrontation with lack – is further theorized through Lacan’s third register, the real.

The subject encounters the real from the place of (the limits of) the symbolic. The real reveals itself as the bodily reminder that something is missing from our social reality, that the symbolic is not an organic whole. ⁵² The real is itself undifferentiated, outside language, inaccessible to signification or representation, and beyond the reach of the subject. ⁵³ Thus the real always emerges as that which resists fixation of meaning. It is the ‘unavoidable thought and somatic experience of that uncertainty, the unattainable fulfillment of the desire to know it’. ⁵⁴ Thus, the real is the affective reminder, for example in the form of anxiety or trauma, that security cannot be fully known, analyzed, cognized, or mastered in terms of security and insecurity.

The Fantasmatic Object of Desire

To avoid confronting the lack in the symbolic and the impossibility of the imaginary unity, which our confrontations with the real reminds us of, a fantasy must be constructed and constantly reconstructed. ⁵⁵ Fantasy gives the subject the means to sustain the illusion that wholeness and security are attainable by concealing the fact that the symbolic ‘is structured around a traumatic impossibility, around something which cannot be symbolized’. ⁵⁶ Thus, fantasy seeks to cover an existential anxiety – that is, an ontological insecurity, or lack, that is an inevitable part of being a subject and signifies the encounter with the real. ⁵⁷ This is why the Lacanian fantasy resonates well with discussions on anxiety within ontological security studies. ⁵⁸ In these studies, the scholarly interest is directed toward the moments in which fantasy fails to manage or repress anxiety. ⁵⁹ When fantasy temporarily succeeds in covering anxiety, it does so by promising a future in which all obstacles and insecurities are eliminated, while also offering specific objects which arouse and reproduce desire for this utopia, but which we do not currently possess. ⁶⁰

This operation ties fantasy to what Lacan calls objet petit a. Objet petit a is the empty position that enables an objectification of lack through different symbolizable objects that promise unity or security when obtained while simultaneously ensuring that the object will never be obtained. ⁶¹ In short, a fantasy is in need of insecurities that explain why full identity, unity, or security is not felt in the present. This is why fantasy relies on the constantly reinvented, always-ambiguous lost objects that cause us to keep desiring. ⁶² Mandelbaum argues that to fully understand the biopolitics studied by many IR scholars, one has to pay more attention to the fantasmatic invention of a promised utopia of homogenous, secure life (congruency between state, nation, and society) hindered by the excesses of another form of life (the foreigner, migrant, Muslim. . .) ⁶³ – also known as the ‘thief of enjoyment’. ⁶⁴

The Economy of Enjoyment

What sustains fantasy, however, is not only the unachievable and ambiguous object of desire. It is also reinforced by a certain economy of enjoyment (jouissance). ⁶⁵ In Lacanian psychoanalysis, the term jouissance is used to describe the fact that language cannot fully represent the subject. Jouissance is the lost fullness, the real enjoyment, and total satisfaction of our desire, which is sacrificed when we enter and, thus, identify through the symbolic system of language, and which we project as stolen by a transgressive, enjoying other. ⁶⁶ While full jouissance remains unattainable, we keep trying to reach it in the margins of socio-linguistics rules, the norms that govern our lives. Thus, as Derek Hook underlines, jouissance becomes a mode of excessive affect, a ‘pleasure in pain’ fueled by the strange gratification of transgressing the conventional everyday-rules of society. ⁶⁷ In fact, the symbolic depends on the jouissance of its subjects to demarcate its boundaries, a reference to something beyond its codified regulations, the Law, in order to function. ⁶⁸

The specific form of enjoyment aroused at the boundaries of Law is also what Žižek refers to as the subjects’ ‘binding of enjoyment’ ⁶⁹ – the kind of thrill you get from indulging in the forbidden, or accusing others of distasteful or unlawful behavior. It is the unsettling, but exhilarating encounter with the illicit, the unauthorized and covert, that binds the enjoyment of the ideological subject, secretly enjoying the feeling of being outside Law, while at the same time emphatically supporting it. ⁷⁰ Thus, the spaces that are off-limit and where transgression of the Law can be enjoyed are not intrinsically subversive; they constitute the ultimate grip of the ideological subject. ⁷¹ As an illustration, Eberle shows how the German opposition to the Iraq war was sustained through tabloid media’s transgressive narratives of the United States as a religious crusader and stupid cowboy, which aroused the necessary affective pull to foster identification with the technocratic, rational war opposition narrative by the German government. ⁷²

The Fantasmatic Reading Strategy – A Three-Step Approach

From the above introduction to the Lacanian tripartite register and its relation to the fantasmatic object of desire and enjoyment, a particular analytical strategy emerges that explicitly conceptualizes fantasy as a negotiation and concealment of lack (unsignified void/absence of meaning/identity/security). From a methodological point of view, this means that security imaginaries are not fully available to discourse analysis. Something will always resist signification and integration in the symbolic order and hence remain outside the grip of discourse. Nevertheless – and this is the central point – it is the never ending attempt and desire to capture the missing (self)identity in discourse, and to cover over the inconsistencies of the symbolic order, that keeps the subject fantasizing. This fantasmatic logic is what affectively drives the security imaginary and makes it stick.

In studying security imaginaries through fantasy, we propose zooming in on three key elements: One, the construction and promotion of a lost utopia that is being prevented by something or somebody (the thief of enjoyment). Two, the investment in and pursuit of a specific object of desire that is constructed to fail. Three, the formation of a particular economy of enjoyment (jouissance) in the encounter with the socially unacceptable, aberrant, and irrational. Each of these steps speak – in different ways – to the study of security imaginaries. First, a historical tracing of a lost utopia (e.g. through policy documents) speaks to the uncovering of a particular security imaginary’s form; that is, its scripts, narratives, and representations of self (and other). Then, locating the object of desire that fills the empty position of objet petit a speaks to the contingent core of meaning constituting the security imaginary. And last, moving to the margins of discourse to capture the subject’s mode of enjoyment speaks to the dimension of affective force sustaining the security imaginary.

This three-step fantasmatic approach to security imaginaries focuses on successful stories – when fantasies are largely successful in covering the anxiety-producing encounters with the real. The reading strategy is thus different from those strategies that put anxiety centre stage, primarily discussed in ontological security studies. ⁷³ Hence, the fantasmatic reading strategy deployed to capture the continuity of security imaginaries does not aim to identify when the real intervenes or what it looks like in terms of anxiety. Instead, it directs attention toward the signifying and affective operations that seek to avoid anxiety-producing confrontations with the real.

In addition to reading strategic documents, the fantasmatic reading strategy should focus on statements and (everyday) practices of key figures, such as policy makers, executives, or bureaucrats as they act as (speaking) subjects engaging in the attempts to stabilize particular scripts, narratives, and representations. In this reading, the subject appears neither as a specific individual or person, nor as an assemblage of institutionalized practices and positions. Rather, our subject of analysis is emerging somewhere in between – thrown in the gap between the imaginary self and the symbolic Other. This provides us with an exemplary case of political subjectivation and negotiation structured around a constitutive lack.

To demonstrate a fantasmatic reading strategy, we turn to a case study of the security imaginary of the US Defense; that is, the narrative of technological revolutions of war. This case study allows us to show what is gained analytically by turning to fantasy, namely that it enables analysts to both reconstruct the form of security imaginary and capture the force with which subjects affectively adhere to and invest in it.

As an empirical starting point we have collected key US policy documents, strategies, and statements on AI-enabled military technology. We have read briefings and transcripts from key US Defense personnel on the issue such as former Deputy Secretary Robert O. Work, and we have trawled through reports and relevant media outlets on defense technology for statements on new military technology, including XAI, by US Defense officials.

Hence, we begin the reading by tracking the promotion of a perceived-to-be lost utopia, using discourse analysis to trace the continuity of specific signifiers. In our case it is the signifier of US military superiority promoted in the US strategic narrative of technological offsets – from the First and Second Offsets in the 1950s and 1970s to the Third Offset Strategy in 2014–8. Secondly, we locate a particular object of desire that exposes the contingent and inconsistent constitution of the security imaginary. In doing this, we follow Glynos and Howarth, when they suggest asking texts whether or not an object resists public official discourse. ⁷⁴ In our study, we locate the US pursuit of XAI displayed in official statements and strategic agendas of US Department of Defense (US DoD). Through texts and speeches, we show that the promise of XAI – as an object of desire – incorporates its own criteria of impossibility, preventing Western militaries from fulfilling the promise of (re)gaining technological superiority. Thirdly, we locate a specific mode of military enjoyment at the margins of Law, a certain affective excess revealed by slips of the tongue in speeches and statements made, in our case study, by key US Defense representatives.

The Fantasy of the US Defense

Tracing the Lost Utopia Through Technological Offsets

Since the victory in WWII, the US military – despite having the biggest budgets in the world – has never felt comfortable in its ability to sustain military domination. The utopia of military superiority always drifts out of reach and new objects of desire that promise to recover the lost utopia are perpetually introduced. The fantasy of recovering this constitutive loss or lack of dominance is articulated in the continuous issuing of military offset strategies that promise to outperform adversary capabilities while simultaneously reducing costs through the development and implementation of advanced and potentially revolutionary technologies.

Starting with the First Offset Strategy in the 1950s, President Eisenhower’s ‘New Look’ strategy was promoted as an attempt to use the science and technology of nuclear power to maintain a strategic advantage over the emerging Soviet Union despite facing increasing budgetary restraints. As a result, nuclear deterrence rather than the more expensive conventional deterrence became the object of desire that promised to maintain or rather regain US military superiority. ⁷⁵ The strategy of investing in nuclear technology to offset the quantitative conventional advantage of the Soviet Union and allow US forces to ‘inflict massive retaliatory damage by offensive striking power’ later became known as the First Offset, although the term was not used at the time. ⁷⁶

Fast-forward 20 years and the failure in Vietnam as well as increasing Warsaw Pact forces in Europe led US Defense Secretary Harold Brown to develop what became known as the Second Offset Strategy. Here, the sense of restraints and disadvantages brought on by the Soviet Union did once again lead to a military strategic reorientation, this time toward a new object of desire prompted by technological innovation, primarily within intelligence, surveillance, reconnaissance, precision guided weapons, stealth technologies, and space-based communication. The US military never tested the offset technologies against Soviet forces, but they proved extremely efficient during the 1991 Operation Desert Storm. The decisive US victory on the battlefields of Kuwait and Iraq meant that the fundamental assumption of the Second Offset Strategy – that superiority in military-technological innovation leads to military superiority – became the axiom of military thinking in the 1990s, particularly through the offshoot-concepts of a Revolution in Military Affairs and Network Centric Warfare (NCW). ⁷⁷

With the rapid advancement of information and communication technology in the late 1990s as well as the realization that superior firepower did not guarantee success against the increasingly insurgent adversaries, NCW’s idea of information superiority now became the key for reaching the lost utopia of US military superiority. For the proponents of NCW, the use of networked information technologies was a way to ‘lift the fog of war’ and let the ‘light of reason shine on the battlefield’. ⁷⁸ By providing near-real-time information, the ‘irrationalities’ of terrain, weather, and human error could be removed and full spectrum dominance based on enhanced situational awareness could be reached. ⁷⁹ ‘While friction and the fog of war can never be eliminated’, the Joint Chiefs of Staff stated in Joint Vision 2010, ‘new technology promises to mitigate their impact’. ⁸⁰ Confronted with the asymmetrical challenges and contingencies in the wars of Iraq and Afghanistan, however, the narrative of NCW lost some of its shine as an object of desire – unable to ‘let the light of reason shine in’ – and deliver the promised utopia of frictionless warfare.

Nevertheless, the fantasmatic narrative of a technological revolution of military affairs persisted. Instead of abandoning the idea of technological offsets as the key to military success, rapid advances in AI and autonomy in the 2010s reignited the quest for an augmented version of networked warfare. In fact, AI and autonomy became the technological backbone of the Third Offset Strategy. The official argument for the need for a Third Offset based on advanced AI and machine learning (ML) systems, was the US inability to overmatch adversary Anti-Access/Area Denial (A2/AD) capabilities of, for example, Russia and China and their growing ability to fight in all domains. ⁸¹ While also focusing on rethinking organizational and operational constructs, the narrative of the Third Offset Strategy was, like its predecessors, sustained by the promise of technological innovations. In a report bearing the title 20YY: Preparing for War in the Robotic Age, former Deputy Secretary Robert Work together with Shawn Brimley, a former director of strategic planning in the Obama administration’s National Security Council, promoted a call to arms, a call that in many ways epitomized the sense of urgency imbued in the narrative of a ‘robotic age of war’. ⁸² In a central passage quoted below, it encapsulated the prototypical fantasy of a military-technological revolution enabled by the proliferation of AI and ML applications:

Since the end of the Cold War, the United States military has enjoyed a virtual monopoly in the guided munitions-battle network regime. . . Now, however, as the United States’ ability to project power and to dominate force-on-force encounters begins to erode as more and more opponents become able to effectively employ guided weapons, defense planners must begin to shift their gaze from the current war-fighting regime to the coming one dominated by proliferated sensors, electric weapons, and ubiquitous unmanned and autonomous systems in all operating domains. . . The United States must overcome this challenge. If it hopes to maintain its technological superiority, the U.S. armed forces must begin to conceptualize how a maturing guided munitions-battle network regime and advances in technologies driven primarily by the civilian sector may coalesce and combine in ways that could spark a new military-technical revolution. It cannot afford to defer the time, thinking and investments needed to prepare for warfare in the Age of Robotics. . . . ⁸³

In this appeal to political and military mobilization and preparation for a robotic age of war, the fantasy of sparking a military-technological revolution is incited by the claim that US military superiority and ability to project power has eroded while strategic competitors have reached technological parity with US and NATO allies. The loss of technological superiority, it is emphatically stressed, can only be recaptured if defense planners begin to invest in a network regime dominated by ubiquitous AI, autonomous, and unmanned systems. In 2018, this strategic narrative was enshrined in the US DoD Artificial Intelligence Strategy: Harnessing AI to Advance Our Security and Prosperity stressing that:

Other nations, particularly China and Russia, are making significant investments in AI for military purposes, including in applications that raise questions regarding international norms and human rights. These investments threaten to erode our technological and operational advantages and destabilize the free and open international order. The United States, together with its allies and partners, must adopt AI to maintain its strategic position, prevail on future battlefields, and safeguard this order. ⁸⁴

Borrowing a Lacanian vocabulary of fantasy and desire, the history and continuity of US technological offset strategies can be seen as a narrative structured around a lack. More specifically, the fantasy of the United States and its allies adopting AI to ‘safeguard the international order’ is sustained by an unending desire to cover the constitutive lack or loss of battlefield superiority, casting China and Russia in the role of transgressive thieves of enjoyment that threaten to erode our technological and operational advantages and destabilize the international order by using questionable applications of military AI.

In short, the above quotes by Robert Works and in US DoD documents promoting an AI-enabled offset strategy illustrate a fundamental experience of loss and insecurity. Something is always lacking, just beyond reach, sustaining a perpetual quest for the elusive object of desire that will, once attained, reestablish security, transparency, and predictability. Hence, in the second part of the fantasmatic reading strategy, we turn to the missing objet petit a – the X of XAI – that allows an endless work of negotiation and justification of why the friction and uncertainties of war are (still) not eliminated. The analytical turn to desire shows that the ‘light of reason’, which is promised to shine from XAI, comes with its own inherent irrationalities and ambiguities, constantly arousing a sense of urgency that affectively engages and mobilizes the subject in the US military imaginary.

The Desire for XAI and Its Condition of Indeterminacy

As the military development and use of high-performance AI and ML algorithms has become ubiquitous and system architectures have become more complex, concerns about transparency and explainability in networked warfare have come to the fore. As an answer to the uncertainties introduced by AI/ML systems – also known as the black box dilemma – XAI is promoted as the technology that will secure lawfulness and an ‘appropriate level of human judgment’ by enabling systems not just to manage huge amounts of data, but also to explain system output and recommendations to human operators. ⁸⁵ Thus, XAI is perceived as the opposite of black box AI/ML systems, ‘ensuring AI advice is trusted, ethical and consistent with national rules-of-engagement’. ⁸⁶ According to the US Army Training and Doctrine Command (TRADOC), this means that ‘artificial intelligence support technologies must be able to explain recommendations, and in the case of autonomous systems provide data, that explains decisions’. ⁸⁷ Similarly, the report from NATO Science and Technology Committee on Artificial Intelligence: Implication for NATO’s Armed Forces underlines that:

AI systems still suffer from severe reliability problems. . . The concept of ‘explainable AI’ and the need for new validation and verification processes specific to AI have thus become critical. ⁸⁸

In fact, nobody seems to disagree, that XAI is needed for the (lawful) conduct of networked warfare. The X in XAI however, remains opaque – oscillating between a plethora of competing legal, ethical, and technical interpretations of when a system is ‘explainable’ since this is entirely dependent on (1) the operational environment, (2) the task performed, and (3) the audience that evaluates it. ⁸⁹ Another problem that remains unresolved is the difficulty of measuring or quantifying whether the details and reasons used to explain recommendations has been understood by the human operator – and whether this makes the use of XAI more ‘ethical’ or ‘responsible’ than other types of AI. Consequently, it has not been determined how predictability and understandability can be consistently and reliably measured, since both are influenced by complex variables like environmental factors and fuzzy characteristics such as ‘human capacity for understanding’. ⁹⁰ The indeterminacy of XAI therefore cannot be resolved by technical means alone, but is intrinsically linked to context sensitive and often competing perceptions of human operators and decision makers.

Due to these obstacles, there is currently no authoritative definition or fixation of XAI. Moreover, the demand for protocols of ‘explainability’ in AI-enabled capabilities will inevitably make the pace of military command and control systems slower, representing a potential disadvantage to systems that are not restrained by these legal and ethical concerns. At a press conference on DoD’s ethical principles on the use of AI, then director of the Joint AI Center Lt. Gen. Jack Shanahan warned about this: ‘I do not believe, sitting here in this room this afternoon, that China and Russia are having any conversation like we’re having today’. ⁹¹ And, ‘the fewer restrictions they have on privacy and civil liberties gives them some advantages in getting data faster, and then building capabilities faster’. ⁹² Former US Army scientist, Rob Albritton presents the argument even more explicitly when he emphasizes that ‘[t]hose kinds of adversaries [Russia and China] have no qualms whatsoever about using fully autonomous weapons whereas our ethics, our morals, don’t allow us to do that – it is just not the American way, not the Western way’. ⁹³ In other words, XAI is an object of desire that comes with its own condition of impossibility: It will – even if legal and technical difficulties are resolved – fail to recover the perceived-to-be-lost utopia of technological superiority since the (unlawful) behavior of the adversaries will continue to give them the advantage of unrestricted warfare.

The intrinsic paradox of the military quest for XAI then, is that it is organized around its own inevitable failure: It sustains the promotion of a lost utopia of security and transparency, while simultaneously providing the reasons needed to explain, why security and transparency still cannot be achieved. This sparks the question: why, in spite of its evident gaps, obstacles, and uncertainties does the fantasy not dissolve itself; why does it persist? What is the appeal of the X in XAI and what does it have to offer? The unavoidable ambiguity and elusiveness of the X – the missing object – must be understood ‘as a way for the subject to organize his enjoyment’, ⁹⁴ a way to ‘choose something instead of nothing’ through the affective investment in a certain signifying object or narrative which assures a minimum of consistency to our being-in-the-world. In this sense, the X becomes that which keeps us from surrendering to the void – the element which ‘gives consistency to the subject’. ⁹⁵

In so far as the X in XAI reveals itself as a particular way of organizing or binding enjoyment that resists determination in discourse, the fantasmatic reading should finally be directed at the uncomfortable gaps and inconsistencies, the excesses, the untimely and the inappropriate outbursts in official discourse (rather than searching for the logical, the socially accepted, the prudent, and the prescribed). It points to the ‘radical dimension of meaningless enjoyment’ ⁹⁶ that cannot be included in discourse, but is at the same time the positive condition of it. This will be further explored in the third and final step of the fantasmatic reading showing how a certain mode of enjoyment (jouissance) sustains the security imaginary of an AI-enabled technological revolution of war.

Enjoying the Margins of Law

The continuous pursuit of (new) objects that are always just out of reach, propels a sense of urgency that cannot be fully captured and stabilized in political discourse. The recovery of the lost utopia is prevented by our own ethical and legal standards and restrictions of warfare, which a transgressive other is not facing (or even manipulating). This is believed to provide him with an (unjust) advantage that must be offset before it is too late. Hence, time is perceived-to-be of the essence and urgent decisions must be made to prevent the adversary from exploiting (enjoying) his transgressions and circumvention of the international order of Law. As former Deputy Secretary of Defense and vice chair of the National Security Commission on Artificial Intelligence, Robert Work puts it:

There has to be a top-down sense of urgency. . . We’re all saying the right things: AI is absolutely important. It’s going to give us an advantage on the battlefield for years to come. But the key thing is, where is our sense of urgency? We may be losing the race, due to our own lack of urgency. ⁹⁷

This sense of urgency spurs an extensive production of policies and organizational initiatives. The avalanche of AI policy documents and institutional reorganizations in DoD is a perfect illustration of this dynamic. ⁹⁸ The former Director of the DoD Joint Artificial Intelligence Center (JAIC), Lt. Gen. Michael S. Groen points to the affective pull created by fear of ‘losing the race’, and of not ‘moving fast enough’ to overcome the obstacles of adopting AI across the DoD:

Are we moving fast enough to create enterprises of capability and overcome stovepiped developments? Are we moving fast enough to really change our operating model to data-driven and data visibility across the department? Are we moving fast enough in integrating innovative technology into the department? And sometimes I lay awake at night and say – and the answer’s no, and that challenge and feeling of hot breath on the back of our necks is what keeps the JAIC motivated and keeps us working hard every day, because we recognize how big this is and the scale of the Department of Defense and how necessary this transformation is at scale. ⁹⁹

The constant effort to overcome obstacles through an excess of new initiatives and extra hard work not only hails the subject to keep identifying with – and thus reproducing – a fantasy. The identification also comes with a mode of jouissance, which cannot be reduced to unbridled enjoyment of lawfulness, a ‘passion for justice’ or the ‘zeal of doing the right thing’. ¹⁰⁰ Rather, it is closely intermingled with the unsettling, yet strangely gratifying identification of a potential space for transgressive practices, suspending the normative and legal prerogatives that are publicly endorsed and defended.

Such an economy of enjoyment is displayed by the morally outraged subject that ‘experiences a giddying rightfulness, a voluptuous contempt’ in excessively condemning those deemed immoral. ¹⁰¹ In the following passage, Robert Work personifies this mode of enjoyment, directing his outrage at the Campaign to Stop Killer Robots (CSKR) – a movement that advocates for a preemptive ban on AI-enabled autonomous weapon systems:

‘They’re [CSKR] defining a weapon that is unsupervised or independent from human direction, unsupervised in its battlefield operations, and self-targeting’, Work said. ‘The weapon doesn’t exist! It might not even be technically feasible, and if it is technically feasible, there’s absolutely no evidence that a western army, certainly the United States, would employ such a weapon’. ‘In the meantime’, Work went on angrily, ‘they’re willing to say, I’m willing to sacrifice the lives of American servicemen and women, I’m willing to take more civilian casualties, and I’m willing to take more collateral damage, on the off chance that sometime in the future this weapon will exist’. ‘That’s unethical to me’, Work said. ‘That’s terribly unethical. In fact, I think it’s immoral’. ¹⁰²

The passage displays an excessive accusation of CSKR. The movement is deemed unethical even though it is motivated by an attempt similar to US DoD policy, namely to secure lawfulness and an appropriate level of human judgment in the use of military AI. While Work may legitimately disagree with the need for a preemptive ban on lethal autonomous weapons systems, it is excessive to call CSKR ‘immoral’ on account of their battle against potentially unlawful AI-applications. In fact, it appears to be an irrational response that is somewhat out of place and out of character for Work as an experienced policy maker. As Berkeley professor and activist Stuart Russell underlines: ‘A more logical response would be for Work to make a counterproposal: treaty language that would allow his “beneficial” applications [of AI] while preventing the creation of WMDs’. ¹⁰³

Work’s agitation is also surprising due to the fact that autonomous computers already control lethal firepower in US missile defense systems, ¹⁰⁴ which is well known and entirely uncontroversial. What remains unsaid by Work, however, is that the current US DoD regulation, Instruction 3000.09 on ‘Autonomy in Weapon Systems’ not only fails to define (or even mention) how to secure ‘appropriate levels of human judgment’, but it also comes with its own ‘legal black hole’, de facto waiving all restrictions on future lethal autonomous weapons systems if only ‘three officials approve’. ¹⁰⁵ Hence, the passion of the morally outraged subject, staged by Work, seems to indicate something more than just the ‘zeal of doing the right thing’ for the United States.

This becomes particularly evident in the DoD efforts of strengthening military–industrial partnerships, cooperating closely with private sector industry: At a hearing in Congress, ¹⁰⁶ the speed and effectiveness of the private sector was celebrated as the key to technological innovation and transformation of the DoD enterprise: ‘[W]e have this vibrant private sector that is able to turn around and take that research and rapidly convert it to commercial products and create new markets’. ¹⁰⁷ This is why the Defense Innovation Unit (DIU) is created and placed in Silicon Valley: to identify promising commercial technologies and rapidly get them into US Defense operations. ¹⁰⁸ During a press briefing, former JAIC Director Groen is even more explicit about the fact that American private sector innovation holds the key to overcoming China’s AI advantages:

[China’s] organizational efficiency, that autocratic rule, they count that as an advantage, is being applied directly to their AI development. . . . Our best opportunities lie in American innovation. Academia and small companies are brimming with good ideas in the AI space. . . . We have the best science and the best AI research available in academia inside the United States and in small companies. . . [and] many of the companies, the modern AI-driven, data-driven companies [. . .] have survived in a very competitive market. ¹⁰⁹

On a closer look however, US public–private partnerships suggest that relying on commercial innovation comes with the potential for the US Defense to enjoy – at least indirectly – those very things that China is criticized for. The first major flagship AI initiative in the US Defense, Project Maven, is illustrative. The initiative is in itself not very controversial. It simply uses AI to help identify potential ISIS fighters in videos recorded by drones. After Google pulled out of the project due to protests from employees, the data analytics company Palantir took over the contract. ¹¹⁰ Palantir is, however, a company surrounded by controversy. The company has been criticized for providing the software that was tied to workplace raids and other controversial actions by the US Immigration Customs and Enforcement agency in violation of due process law, as well as helping local police forces gain access to suspect’s everyday life pattern without a warrant. ¹¹¹

While such human rights violating surveillance practices mirror what the Chinese government uses AI for, nowhere has the US DoD mentioned the controversies in relation to their partnership with Palantir. Such omissions from speeches or official documents are interesting as they suggest that there is a repressed or disavowed desire to enjoy a space beyond the boundaries of Law and to suspend the institutionalized norms and values that are declared as the guiding principles of the US DoD. Such repressed desire in the US Defense supports arguments made in surveillance studies that the US representations of China as an excessive thief of enjoyment are discursive operations that give form to the intelligence community’s own transgressive desire. ¹¹²

Once again, this is illustrated by Work – pushing for a so called ‘Naval Reactor model of organization’ of the DoD’s JAIC, a model designed in the 1950s by Adm. Hyman Rickover, allowing extraordinary independence from both military and civilian oversight. ¹¹³ According to Work, the classic bureaucratic organization of the DoD is simply unable to keep up with competitors such as China, which has the benefit of a single unified Strategic Support Force. ¹¹⁴ As Freedberg comments, the idea of replicating such a model ‘would upset rice bowls in the Defense Department and industry alike, making reform an uphill battle both politically and bureaucratically’. ¹¹⁵ In his own words, Work would hope ‘that the department really goes after bold changes, and not let the bureaucratic inertia of the department hold it back’. ¹¹⁶

When Work was later confronted with his bureaucracy-bursting model of organization, his reaction was a gleeful dismissal: ‘God, did I really say that? . . . So right now, I would say I’ve changed from a Navy Reactor model to the National Security Commission on Artificial Intelligence model’. ¹¹⁷ Instead of denying his political appeal to override the bureaucratic inertia of the department, that is, the rational and rule-based governance of the military apparatus, he succeeds in sharing his desire to enjoy freedom of bureaucratic restraint with his audience – an unspoken, but effective hailing of political subjectivity at the margins of Law.

Conclusion

The study of security imaginaries through the Lacanian concept of fantasy helps us to better understand how certain narratives, myths, scripts, etc. are sustained and reproduced – not only because they have status as common sense or background knowledge, but also because of the subject’s affective investments. The article introduced a three-step fantasmatic reading strategy that embraces both the discourse analytical unpacking of ‘the common sense’ of a security imaginary and the affective investment that captures the force with which subjects continue to invest in the security imaginary. The former, we suggest, can be done through the tracing in security discourses of a lost utopia that is being prevented by something or somebody (the thief of enjoyment). The latter can be achieved, firstly, through the tracing of the subject’s pursuit of a particular object of desire that comes with its own condition of impossibility, and thus inevitably will fail to recover the perceived-to-be-lost utopia. And second, the affective investment should be analyzed through the tracing of the particular mode of enjoyment that emerges in the encounter with the socially unacceptable and irrational. The article explored the analytical insights gained from utilizing this three-step approach in a study of US Defense narrative of ‘technological revolutions of war’.

The article’s Lacanian starting point formed our theoretical assumption that a security imaginary is structured around a constitutive lack (of meaning and security) articulated as an immanent loss or erosion of an identity position – which in our case study is US military superiority. Because the US narrative of a technological revolution of war is constituted by an irreducible lack – an original trauma – a fantasy must be constructed promising to repair lack and reinstate the perceived-to-be-lost (though never achieved) state of security and predictability. In a fantasmatic reading, what comes to the fore is the endless work of covering the absence of stable identities and positions, drawing subjects into continuous negotiations of in/security and un/certainty.

This fantasmatic approach allows us to take a critical stance to the ongoing efforts to build common AI policy in the international community. Rather than building on the premise that explainability and transparency in advanced AI/ML systems is within reach of legal and technical amendments, it may be more appropriate to assume that a constitutive lack of explainability and transparency is an inherent and unavoidable characteristic of networked AI/ML capabilities. By this, we do not in any way endorse an ‘anything goes’ approach to military AI. On the contrary, a Lacanian approach to military imaginaries points to the need for a more embedded practice of jurisprudence in the entire life cycle of military AI design, acquisition, and implementation allowing for continuous evaluation and negotiation of emerging vulnerabilities and in/securities. Hence, we suggest, political attention should be directed toward emergent areas of illicit and transgressive uses of military AI. In particular, increased scrutiny of the hybrid and distributed networks of private–public partnerships is required, where legal responsibility and liability in the event of system malfunction or unlawful conduct can be difficult to place. In a Lacanian reading, such blind spots pull the subject of insecurity into excessive and often counterproductive attempts of covering the gaps and irregularities of war.