Seeing, Knowing, and Deciding: The Technological Command Dream That Never Dies?

Could Gen. William Westmoreland see the longer term? In a 1970 situation of Army Aviation Digest, Westmoreland, then chief of employees of the U.S. Army, supplied his view on the way forward for command decision-making. In the print model of a speech he had delivered the yr earlier than, he predicted the observe of command would happen on a extremely surveilled, automated, and interconnected battlefield which, if such a system have been realized, might “assist the tactical commander in making sound and timely decisions” in addition to allowing “commanders to be continually aware of the entire battlefield panorama.” Although Westmoreland was definitely not a clairvoyant, in discussions of up to date desires of command his language wouldn’t seem misplaced. This is particularly the case surrounding efforts to implement Joint All Domain Command and Control, which, in broad phrases, is the U.S. navy’s effort to hyperlink decision-makers throughout land, sea, air, area, and cyber utilizing superior communications and knowledge evaluation applied sciences.

The launch of the Department of Defense’s Joint All Domain Command and Control technique in early 2022 supplied some extra data relating to the objectives of the division’s supposed reimagining of the technological components supporting command within the U.S. navy. The doc is couched within the rhetoric of great, technologically impressed modifications that may enable the navy to research and course of knowledge from a variety of sources in addition to resolve sooner, based mostly on superior evaluation capabilities. According to the declassified abstract, the technique is structured round three pillars: “sense,” “make sense,” and “act.” The technique’s drafters envision these core elements being enabled by the knowledge and decision-support capabilities of synthetic intelligence, machine studying, and superior sensor techniques, of which “information and decision advantage” on the “speed of relevance” is meant to be the consequence. Westmoreland’s “sound and timely” choices, in addition to his notion of a commander being conscious of the “entire battlefield panorama” would match effectively into this schema. So too would Adm. William Owens’ want to “lift the fog of war” by means of the mixing of recent data and communication applied sciences 30 years later.



As articulated within the technique, such modifications are proposed as a response to novel threats. Yet, as each Westmoreland and Owens’s views attest, a evaluate of previous imaginations of command-related applied sciences within the U.S. navy makes one factor clear: The imaginative and prescient for a up to date superior command system is yet one more instantiation of a longstanding dream associated to the orchestration of navy choices, slightly than one thing fully authentic. In such visions, the U.S. navy dangers what B.A. Friedman and Olivia A. Garard counsel is the tendency to conflate “technological capacity with command.” Thus, in mild of the current launch of the technique, it’s value exploring the rhetoric that sustains these reoccurring visions of technologically enabled command. Furthermore, we must always contemplate the sensible implications of technological roadblocks dealing with present efforts in addition to attainable perils associated to its guarantees of enabling sooner and higher choices.

Echoes from the Past

While such visions have roots that hint at the very least to missile protection applications just like the Semi-Automatic Ground Environment within the Nineteen Fifties — in addition to early Pentagon-supported computing analysis initiatives for functions of command and management in the course of the Sixties — notions of command techniques approximating the present intentions start to coalesce extra firmly in the course of the latter many years of the Cold War. We can contact on just a few examples that resonate with present efforts.

Driven by navy competitors between the Soviet Union and the United States, in addition to fears over the implication of Japanese fifth-generation computing, the Defense Advanced Research Projects Agency undertook a decade-long, billion-dollar effort generally known as the Strategic Computing Initiative. Starting within the early Nineteen Eighties, it featured a multi-pronged effort to discover the implications of AI for navy functions — a key ingredient of which was command and management. This got here notably within the type of a proposed battle administration system for the U.S. Navy. Like present efforts, Strategic Computing’s contribution to command-related analysis talked about AI-enabled knowledgeable techniques serving to commanders rapidly sense and resolve what to do. The program’s founding doc famous the aim of growing human-like, “intelligent capabilities for planning and reasoning,” the need to enhance human judgment with AI-enabled knowledgeable techniques, and the necessity to assist make sense of “unpredictable” navy conditions in a speedy trend. The plan was to create a decision-support system that went past current choices in the course of the period. The principal possibility on the time was the Worldwide Military Command and Control System, itself predicated on “accurate and timely decisions”, by leveraging clever computing that would help in planning, formulating determination choices, and managing uncertainty in rapidly altering fight environments. In the top, as Emma Salisbury factors out, this system had combined outcomes and usually did not dwell as much as its large AI-related guarantees on account of technological roadblocks in addition to funding shortfalls.

Yet nearly as quickly as Strategic Computing fizzled out, related high-tech approaches to command techniques popped up once more. In the late Nineties, the U.S. Army’s Future Combat System proposed a modernized command construction linking manned and unmanned techniques over wi-fi networks. The initiative emerged out of assumptions about battle related to the Revolution in Military Affairs, wherein protection officers believed smaller, sooner, forces linked by superior communications networks would show decisive in future conflicts. While the modernization effort additionally supposed to interchange some navy {hardware} such because the M1 Abrams tank, it was the command community that bonded this system collectively. The want for velocity and the notion that data dominance may show crucial in future conflicts drove the Future Combat System. In phrases of command, as specified by a 2007 Congressional Research Service report, this manifested in technological artifacts resembling ‘Battle Command Software’, the need for automated mission planning capabilities for speedy response, in addition to the intent to enhance ”scenario understanding” by means of using maps and databases that would monitor enemy areas. In reality, throughout congressional budgetary hearings, protection officers cited the supposed benefits incurred by troopers throughout testing, principally stemming from “increased soldier awareness and battlefield understanding.”

Similar to Strategic Computing, the Future Combat System endeavored to assist commanders assess, make sense of knowledge, and act faster than was beforehand attainable. However, additionally akin to Strategic Computing, this system struggled to dwell as much as its guarantees. A 2012 RAND evaluation of this system largely thought of the undertaking a failure due, at the very least partly, to over-aggressive timelines and shifting aim posts. While RAND’s report on the Future Combat System solely took on facets of command and management in a restricted trend, it did level to efficiency points within the command system’s skill to finish duties resembling automated knowledge fusion to generate a typical operational image. The report famous that such points degraded “a key operational linchpin” of this system.

Although the Future Combat System and ongoing plans to develop a high-tech command system aren’t exactly the identical, the parallels between the initiatives are at the very least shut sufficient for protection officers to contend that present efforts don’t repeat previous errors. Even amongst such claims, there are notable similarities between the Future Combat System’s intentions to assist the Army “see first, understand first, act first” and modern efforts to “sense, make sense, and act” at a faster tempo.

Apparently, desires of technologically enabled sensemaking and quick choices die onerous. Ongoing justifications of recent technologically superior command techniques generally depend on rhetoric that coheres with the wishes of its predecessors in each Strategic Computing and the Future Combat System. For instance, members of the Joint All Domain Command and Control improvement crew have argued that battle sooner or later will function determination timelines which might be lowered to “milliseconds” necessitating the mixing of developments in computing, AI, and machine studying because the “linchpin” of command necessities. As the technique argues, such technological modifications are essential for processing and delivering data on the speeds required in trendy battle.

Furthermore, with respect to AI and the technological components supporting navy decision-making, related rhetoric is deployed outdoors of discussions particular to present command visions. The 2021 National Security Commission’s Report on Artificial Intelligence, co-chaired by former Deputy Secretary of Defense Robert Work, asserts that, in navy contexts, AI will assist to “find needles in haystacks” and “enhance all-domain awareness” resulting in “tighter and more informed decision cycles.” Similarly, the Defense Advanced Research Projects Agency’s notion of Mosaic Warfare is itself a metaphor for the intent to dynamically hyperlink weapons, sensors, and decision-makers. Such arguments seem throughout us and mirror the diploma of technological optimism surrounding AI-enabled techniques. Interestingly, whereas these reoccurring wishes for technologically enabled command techniques are framed by way of “rapid changes” to the safety ecosystems and “significant new challenges” dealing with the United States, what appears to be occurring is that the Department of Defense has turned once more to its decades-long aspiration. For justification, it leans on rhetorical assertions and problematic assumptions that even have decades-long histories. And whereas strategic and political elements have modified within the time spanning from the Cold War to modern safety challenges, related tech-optimistic framings persist.

Inertia and Implications

Today, Joint All Domain Command and Control definitely has a level of institutional momentum. As Gen. John Murray (now retired) acknowledged in a 2020 congressional listening to, “nobody is arguing with the concept.” Furthermore, as just lately as this yr, Gen. Mark Milley proclaimed the “irreversible momentum” towards this system’s implementation. Whether that momentum pushes in a coherent route or not is unclear, even contemplating the discharge of the current technique. What must be famous, nonetheless, is that within the days since initiatives resembling Strategic Computing and the Future Combat System, AI/machine studying, the {hardware} and knowledge availability wanted to coach algorithms, and different digital applied sciences have develop into extra superior and succesful. Nonetheless, AI/machine studying continues to be stricken by critical issues resembling bias, points with coaching knowledge, belief in machine-human interplay, and difficulties associated to explainability, or the need to know why an AI system got here to the choices it did, although the latter is a matter the Department of Defense is engaged on. As an instance, as prompt in a 2022 report by Stanford’s Human-Centered Artificial Intelligence, even in pristine analysis environments of universities and personal sector corporations, the place datasets may be labeled and curated and algorithms skilled and examined, there are nonetheless issues associated to language fashions reproducing the bias of their coaching datasets at more and more excessive charges. Thus, with respect to navy AI, the issue is that even in excellent settings machine studying fashions could make surprising and undesirable errors.

Furthermore, many fashions are skilled with efficiency on benchmarking datasets as the last word aim, not their performance in the true world. As machine studying researchers have prompt, it’s because fashions are continuously “taught to the test” of the benchmark knowledge, resulting in outputs that may’t be replicated outdoors of managed analysis environments. Therefore, main questions persist in relation to AI/machine studying’s place in navy command processes. For instance, what knowledge will the weather of command techniques counting on the prediction and determination assist capabilities of AI be skilled on? Additionally, how sure ought to officers be of system efficiency within the complexity of battle, notably if AI-enabled command techniques are providing up suggestions or attainable programs of motion? Avi Goldfarb and Jon Lindsay have just lately pointed to the problems with knowledge in navy environments. There is a significant danger to not fixing such issues, notably within the case of AI-enabled command choices.

That mentioned, time horizons matter. Current efforts surrounding Joint All Domain Command and Control are principally centered on initiatives resembling cloud capabilities and enhancing interoperability and knowledge sharing throughout the providers and different companions. The implementation plan stays labeled, so efforts associated to utilizing AI for prediction or planning are murky. Still, it’s value assessing what the navy could flip to subsequent within the context of the current public technique doc. Particularly if AI-enabled techniques are envisioned as offering, because the technique suggests, the “technical means to perceive, understand, and predict the actions and intentions of adversaries, and take action.”

Apart from issues relating to technological performance, we must also assess what the dominant assumptions contributing to present efforts may imply in sensible phrases. As addressed above, the longing to behave sooner and scale back confusion on the battlefield are outdated wishes in navy thought, ones which might be continuously linked to the supposed capabilities of superior computational techniques. The implication of those long-term issues — if they’re thought of solvable by means of command-related applied sciences — is worrying. This is especially the case if such applied sciences are envisioned as the trail to attaining a speedy, decisive victory.

Scholars have documented the perils of assuming, and pursuing quick, decisive battle. In reality, Dave Johnson just lately explored this situation in War on the Rocks, wherein he touches on and critiques the “belief that future wars will be short, decisive affairs.” Furthermore, Antoine Bousquet argues that science and expertise have been repeatedly turned to by militaries searching for such decisiveness. While it is very important not conflate basic strategic issues and tactical choices on the battlefield, the choices commanders make ought to assist total political goals. And as Paul Brister suggests in a current Brookings edited quantity, there are robust classes to be realized from assuming technologically enabled tactical velocity will result in brief or simply received wars. In the identical quantity, Nina Kollars argues a parallel level suggesting, “the lure of faster war leading to faster victory is not only questionable but also a persistent techno-pathological obsession.” Thus, speedy, AI-enabled choices as a method to rising tactical velocity shouldn’t be seen as undeniably helpful, notably within the face of present discussions of so-called “hyperwar.”

Notably, Joint All Domain Command and Control by itself just isn’t a idea of victory or a warfighting idea. And it doesn’t advocate a speedy finish to battle as such. Rather, its proponents envision it as an enabler of different warfighting capabilities by means of establishing, at the very least in its preliminary phases, a extra interoperable set of knowledge techniques. The proposed results of such endeavors is, because the technique contests, to “directly and dramatically improve a commander’s ability to gain and maintain information and decision advantage.” That mentioned, its proponents danger forwarding superior command techniques because the enablers of speedy — decisive — victory, thus overly shopping for into technological optimism as an answer to attaining political or strategic objectives. As the imaginative and prescient for a brand new command system continues to emerge, we’d be higher served by being cautious of any “techno-pathological obsession.”

Furthermore, versus some conclusions from the National Security Commission on AI, it isn’t fully obvious that AI-enabled techniques shall be profitable in clarifying very a lot within the conduct of warfare. As Sam Tangredi notes, AI techniques nonetheless wrestle with issues that aren’t well-structured, and battle is something however a well-structured drawback. Moreover, deep-learning AI fashions have confirmed to be comparatively simple to trick by means of, for instance, altering pixels in photos fed to the algorithm. Significant components of battle are associated to diversionary ways and stratagems meant to confuse and mislead adversaries. This mixture of a technical drawback and the frequent observe of deception in battle could result in additional confusion slightly than readability for these working inside any totally realized model of an AI-enabled command system.


Accordingly, we must always hesitate to conclude that new command-related applied sciences will all of a sudden illuminate the battlefield for commanders resulting in quickly executed higher choices, or essentially result in anticipated constructive navy outcomes. Further, we must be skeptical that such wishes are all that new, and, as Martin Van Creveld’s work on command suggests, they may stay very onerous to good. Justifications for growing a brand new, extra superior, command system echo previous tasks such because the Strategic Computing Initiative and the Future Combat System, each of which have been usually unsuccessful in hitting their bold marks. It is vital to at the very least contemplate the rhetoric, and the outcomes, of those previous tasks in debates over the deserves and prospects of future technologically superior command techniques.

Furthermore, the rhetoric surrounding the event of recent high-tech command techniques is doubtlessly dangerous whether it is substantively linked to assumptions about fast-acting, technologically decisive battle. There are historic instances which show the results of instantiating related assumptions into navy observe. Azar Gat’s work on the historical past of navy thought main as much as WWII, whereas too advanced to delve into element right here, demonstrates that myth-like concepts about technological artifacts resembling motor automobiles and airplanes formed how theories of quick, mechanized battle have been put into use. Thus, a reflective consideration of the historic similarities associated to linking quick choices and enhanced situational consciousness with superior computation or AI-enabled techniques, in addition to foregrounding the very actual technological hurdles dealing with the mixing of AI-related applied sciences into determination processes, will assist to avert the worst outcomes.



Ian Reynolds is a Ph.D. candidate in International Relations on the American University School of International Service learning the historical past and cultural politics of navy synthetic intelligence. He can also be a doctoral analysis fellow on the Internet Governance Lab and a analysis affiliate on the Center for Security Innovation and New Technology, each housed at American University. During the 2022/23 educational yr, Ian shall be a pre-doctoral fellow at Stanford’s Center for International Security and Cooperation in addition to the Institute for Human Centered Artificial Intelligence. 

Image: U.S. Army


Be the first to comment

Leave a Reply

Your email address will not be published.