The speed of the attack quickly overwhelmed nearly all the ship’s combat systems, and while the information technology specialists were able to release some defensive systems from the clutches of the cyber intrusion, the sailors in the combat information center (CIC) simply were unable to generate the speed to react. Decision-action times were in seconds or less. Indeed, it appeared from the now very limited situational awareness in the CIC that some of the enemy autonomous weapons were providing support to other systems to set up attacks of other systems. The entire event was over in minutes.
The captain had survived, courageously remaining on the bridge, but he was badly wounded, as were many crew members. Fires were burning out of control, and the ship was listing badly from flooding. Because of the damage, the captain was unable to communicate to the damage control assistant (DCA), who was, herself, badly wounded but valiantly seeking to control the fires and flooding. Damage control central had been hit. Evidently some of the autonomous platforms knew exactly where to strike the ship to both maximize damage and reduce the chances of survivability. With his capacity to command the ship now badly compromised and the flooding out of control, the captain did what no U.S. skipper had done for generations—he issued the order to abandon ship.
On only a few occasions has history witnessed fundamentally transformative changes in the way war is waged. The employment of cavalry, the advent of the rifled musket, and the combination of fast armor with air support and instantaneous radio communications in the execution of the Blitzkrieg strategy are a few examples. Technological developments—sometimes originating in a variety of different fields—come together to enable these seismic shifts. Another such shift is coming soon to the field of battle. Those who are not prepared for it will fare no better than the Iraqi Army did when confronted with the “second offset” technologies of smart, precision-guided weapons, stealth, and electronic warfare.
Broad contours of how this new shift in the way war will be waged already are becoming clear. Technologies such as computer vision aided by machine-learning algorithms, artificial intelligence (AI)-powered autonomous decision making, advanced sensors, miniaturized high- powered computing capacity deployed at the “edge,” high-speed networks, offensive and defensive cyber capabilities, and a host of AI-enabled techniques such as autonomous swarming and cognitive analysis of sensor data will be at the heart of this revolution. The major effect/result of all these capabilities coming together will be an innovation warfare has never seen before: the minimization of human decision making in the vast majority of processes traditionally required to wage war. This minimization likely will alter where the human will be located in the decision-action loop and the human’s specific involvement in decision making itself. In this coming age of hyperwar, we will see humans providing broad, high-level inputs while machines do the planning, executing, and adapting to the reality of the mission and take on the burden of thousands of individual decisions with no additional input.
First, why refer to this AI-fueled, machine-waged conflict as hyperwar? This is not a new term. In World War II, its use implied the global nature and many concurrent theaters of war. In today’s context, however, hyperwar may very well be applied globally, but the element of “pan-war” is not its singular defining characteristic. Instead, what makes this new form of warfare unique is the unparalleled speed enabled by automating decision making and the concurrency of action that will become possible by leveraging artificial intelligence and machine cognition.
In describing the wars of the future, “hyper” is used in the original Greek sense of the word—“over” or “above.” This new type of combat will be beyond what has been seen before in important ways. In military terms, hyperwar may be redefined as a type of conflict where human decision making is almost entirely absent from the observe-orient-decide-act (OODA) loop. As a consequence, the time associated with an OODA cycle will be reduced to near-instantaneous responses. The implications of these developments are many and game changing.
Infinite, Distributed Command & Control Capacity: Until the present time, a decision to act depended on human cognition. With autonomous decision making, this will not be the case. While human decision making is potent, it also has limitations in terms of speed, attention, and diligence. For example, there is a limit to how quickly humans can arrive at a decision, and there is no avoiding the “cognitive burden” of making each decision. There is a limit to how fast and how many decisions can be made before a human requires rest and replenishment to restore higher cognitive faculties.
This phenomenon has been studied in detail by psychologist Daniel Kahneman, who showed that a simple factor such as the lack of glucose could cause judges—expert decision makers—to incorrectly adjudicate appeals. Tired brains cannot carefully deliberate; instead, they revert to instinctive “fast thinking,” creating the potential for error. Machines do not suffer from these limitations. And to the extent that machine intelligence is embodied as easily replicated software, often running on inexpensive hardware, it can be deployed at scales sufficient to essentially enable an infinite supply of tactical, operational, and strategic decision making.
Concurrency of Action/Perfect Coordination: “Overpowering the enemy” is a phrase used often in the literature of war. In military terms, this refers to the concentration of force in a finite space, over a finite period of time, such that the application of this force against the opposing elements able to respond delivers a numeric or firepower advantage impossible for the opposition to counter or resist. This may not necessarily be because the attacking force is larger or more powerful than the entire defending force, only that it is more powerful when and where it matters. This is an important distinction. If a smaller force can be quickly “perfectly coordinated” and applied to a precise point where the enemy is unable to reinforce over the period of hostility, then the smaller force usually will prevail. If such action can be replicated repeatedly, then much larger opposing forces can be effectively neutralized economically and often will be dislocated psychologically.
The two key variables of concern are time and space. The time is what it takes to form and execute kinetic action, and the space is where such action is to be executed. These variables are computed as a result of significant strategic, operational, and tactical decision making. Identifying a candidate space for the application of force is the first ingredient. When done properly, it involves computing a large set of contingencies, called branches and sequels in planning parlance, regarding the enemy’s capacity to replenish, resupply, and reinforce. The tactical matters of identifying targets, maneuvering to achieve advantage or to avoid counter fire, and directing one’s own fire add to this list of decisions and to the cognitive complexity. With machine-based decision making, a large group of sensors and shooters can be coordinated instantaneously, enabling the rapid forming or massing of forces and the execution of kinetic action and subsequent dispersal.
The degree to which concurrency of action can be achieved with machine-based decision making fuels hyperwar and will far outpace what can be done under human control and direction.
Logistical Simplification: The old adage that, “amateurs talk tactics, and professionals discuss logistics” is good guidance. Since time immemorial, waging war has required the movement of human armies that must be fed, clothed, and protected. When the level of intelligence required to fulfill a specific mission can be created in synthetic form, however, machines can become soldiers. The needs and logistics demands of robotic soldiers will not be as varied as those of a human soldier, nor will these machines be as indispensable as a human soldier. The loss of these assets no longer will trigger the expensive and dangerous standard operating procedures involving infiltration of a medical team, extraction, and transportation to a field facility.
Today’s drones or unmanned combat aerial vehicles (UCAVs) mostly are remotely piloted systems that simply separate the human pilot from the craft, placing human decision making at a distance. This is a useful configuration, but it has many downsides. First, the latencies involved mean that only certain types of missions can be fulfilled by today’s drones. High-speed air-to-air combat would be difficult, for example. Second, the system remains susceptible to jamming and loss of communications. Third, the human pilot succumbs to many of the pressures and stresses of real war. This drone pilot post traumatic stress disorder phenomenon has been well documented and sheds light on the limitations of the current model.
Truly autonomous UCAVs of a variety of types and sizes with on-board synthetic intelligence will be the foot soldiers in a future hyperwar. Models the size of commercial quadcopters capable of weaving through forests and racing across open fields will assemble, act, and dissipate in no time. They will be armed with sophisticated sensors that feed vision and decision-making algorithms both on board, in the swarm, and when accessible, in centralized locations. In addition, they will come equipped with a variety of cyber and kinetic payloads. A large number of these systems can be coordinated by means of swarm algorithms, enabling “a collective” to ensure the fulfillment of a mission and individual drones to support and to adapt to the loss of another.
Despite their flexibility, these systems principally will require only two resources: energy and ammunition. In the future, energy may be converted to ammunition, such as with directed-energy weapons. Still, it will be some time before the requisite miniaturization can be achieved to deliver this capability. These assets will remain “resource neutral” until they are actively being employed, reducing the overall energy required to sustain them in a theater over time. With all these changes, the logistical effort will be simplified immensely, and as a result, the “teeth to tail” ratios for autonomous forces will be higher than for any manned force.
Instant Mission Adaptations: German World War II General Erwin Rommel once said, “The best form of welfare for the troops is first-rate training.” Without training, there is no chance of success, and advanced forms of military training help create specializations for roles that are essential in the conduct of war. In the face of artificially intelligent technologies and the hyperwar they will enable, there will be two groundbreaking changes in training.
First, AI technologies such as natural language-based dialog systems that can ingest hundreds of thousands of pages of manuals, guides, studies, and more will augment human operators in non-combat situations, such as with maintenance and remediation of equipment. Eventually these capabilities will be enriched with augmented reality information-delivery technologies in combat scenarios.
Second, when employed in an entirely autonomous fashion, the tactics and strategies of an AI system—its entire set of behaviors and corpus of acquired knowledge—can be copied easily from one system to another. This is the equivalent of having the most qualified veteran instantly transfer his or her experience and expertise to troops who have never been in battle. Further, an AI system’s skills and specializations can be swapped in and out immediately. The same autonomous aerial platform can be an expert “pilot” for a suppression of enemy air defenses mission and, with a quick swapping of the neural network controller, become the world’s deadliest air superiority specialist. In addition, if one such “expert” AI pilot needs to be sacrificed to achieve mission objectives, so be it. Other than the hardware, nothing is lost. The “brains” of the pilot simply can be replicated on a different piece of hardware.
Training for AI-based systems can happen in the real world, or in simulators. An approach known as “reinforcement learning” has made great strides in defeating human players at traditionally unconquerable games, such as the ancient game of Go. The same technology is being employed to build better autonomous cars. Each autonomous car does not have to go through the learning curve that every human driver must navigate. Instead, the car—or simulated car—that evolves the best performing neural network can communicate that experience and learning instantly to all other vehicles. This instant “transfer learning” will be another unparalleled reality in future hyperwar, fueled by the employment of artificial intelligence.
28 May 2027
An Autonomous Defense Rises
The artificially intelligent cyber defense system in the guided-missile destroyer’s CIC was the first to detect what appeared to be an attempt at a major cyber intrusion, perhaps an attack. The intrusion was pervasive, seeking to lock-out the ship’s sensors and many of its defensive systems, and seemingly concentrating on the ship’s antiswarm batteries (ASBs) and supporting systems. The initial cyber attack and the successful defense occurred within microseconds. The defensive system had functioned exactly as it had been designed. As a result, the ship was able to “sense,” then detect, a massive incoming complex swarm attack—the kinetic follow-up to the invisible opening strike. In fact, the system had gone further, instantly forwarding threat information to the rest of the fleet, enabling other units to prepare for an impending attack.
The captain moved quickly from the bridge into the CIC and, along with the others in the center, donned the augmented reality headgear and attendant gauntlets to assimilate and react to the totality and complexity of the battle he was about to lead. His first thought was the status of his weapons. He had only seconds as some elements of the swarm were supersonic, maybe hypersonic. Because of the elevated threat level, the captain had been given a high level of authority and autonomy to engage any potential attackers. He quickly cycled to the “weapons status” views in his headset, and all were green, being continuously fed targeting information from the ship’s fire-control complex now locked onto and tracking and analyzing the incoming attacking swarm. He had to act and shifted to the “ASB status view.” With a sweep of his hand in virtual reality, he initiated the ASB.
In that instant, naval warfare changed forever. Now, “cleared hot,” the various components of the ASB sprinted skyward outside the skin of the ship, and the airspace was filled with several types of now-completely autonomous aerial vehicles. Some moved off at high speed on the azimuth of the incoming attack to engage the enemy swarm at long range; others dwelt in the vicinity of the ship, ready to engage as a last-ditch defense. No one on the ship, indeed no one in the U.S. Navy, had experienced the ASB going into action at full capacity. The ship shuddered as systems leapt into the air with a cacophony of noise.
Back in the CIC, the captain shifted to “target view” in his headset to see what was coming. He had been slightly skeptical this would all come together, beyond his simulator training, but now he was seeing the reality of something nearly beyond belief: completely autonomous aerial systems locked in mortal combat. Blue tracks representing ASB systems and red tracks identifying enemy threats filled the screens. Likely electronic countermeasures (ECM) decoys were highlighted in orange and automatically deprioritized by the ASB. As the battle unfolded—measured in seconds—one after another red and blue systems winked out as they crashed into each other or detonated in close proximity. That battle was moving toward his ship at a high speed. Having donned his own headset, the weapons officer quickly unleashed the full might of the various close-in weapon systems, including the autonomous systems from the ASB, which continued to engage the closing enemy swarm.
The first impact was deafening. Some elements of the enemy’s swarm had detonated above the ship, taking out some of the ship’s antennas. They evidently were searching for certain antennas to reduce the ship’s connectivity. The second strike carried away a 20-mm. Phalanx Gatling gun, a principal means to defend the ship. The third blast struck the ship at the water line, killing and wounding a number of crew members and starting fires and flooding. While outside the ship a maelstrom was unfolding as kinetic systems autonomously coordinated fires with the near continuous launching of the ASB, inside the ship, damage-control and medical recovery measures were under way.
The captain quickly switched to “damage control view” and was able to see the AI-enabled dashboard view of the damage and the damage-control measures the ship’s DCA was using to fight fires and control flooding. Because of the sophistication of the AI system, he could instantly “see” which of the ship’s systems were offline, which were being rebooted to recover, and which were being instantly cross-connected to restore capacity and capability. The AI-powered damage-control system was quickly and autonomously shifting power loads and bringing emergency systems on line. Decisions for damage control were being made in seconds where before long minutes were needed.
The captain then shifted to the view he dreaded: “crew status.” Because every member of the ship’s company wore a “health status harness,” which measured body temperature, heart rate, blood pressure, and breathing, he instantly could see the overall status of his crew and each individual sailor’s status dashboard. Sobered and saddened by the number of casualties as he cycled through subviews in this domain, he saw who had been killed and who was wounded. He knew which of his leaders were down and began to consider how he would reconstitute the chain of command.
Hours later, with his wounded cared for, the fires out, and the flooding under control, the captain reflected on the engagement. He was shaken but not frightened by the reality. The attack had come seemingly from nowhere. The cyber defense system had detected the initial cyber intrusion, and not only had it protected the ship, but it also had reasoned the attack was a precursor to something larger and alerted the CIC of what might be coming. This hypothesis had been formed, researched, and validated in less than a second. Within ten seconds, the ship initiated general quarters on its own and the captain had donned his augmented-reality ensemble. From that moment until the final fires were put out, using the automatic fire suppression system coordinated around crew status readings, the entire battle had unfolded and was over in minutes.
The autonomous nature of the ASB assets, coordinated with the CIC, and the ship’s defensive systems had foiled a coordinated, complex cyber and autonomous swarm attack. The captain was struck by the realization that at nearly every point where human actions and decisions were required they nearly risked the ship. Though he was a master of the combat systems of the USS Infinity (DDG-500), he had just experienced the near mind-numbing speeds of AI and deep-learning-driven warfare. He had become the first U.S. commander to fight in the environment of hyperwar.
Is This A Revolution in Military Affairs?
The scenarios here and the intervening discussion provide a window into only a few of the ways in which synthetic intelligence will fuel the next great shift in how warfare is conducted. The fusion of distributed machine intelligence with highly mobile platforms brings a speed and scale of concurrency never seen before. The hyperwar these technologies will enable is a new paradigm for which we need to plan. The rise of these capabilities has sparked a revolution. But it is more than a revolution in military affairs, it is a revolution in human affairs with major implications for the security and defense arenas. Advances in AI have the capability to fundamentally change the human condition, and with it, a profoundly human undertaking, war.
Near-peer opponents already are investing heavily in these technologies and have some operational AI-powered weapon systems, such as cruise missiles. The ability for autonomous algorithms to transform moderately dangerous weapon systems into significant threats means we must watch for and guard against synthetic intelligence being added to existing arsenals.
The speed of battle at the tactical end of the warfare spectrum will accelerate enormously, collapsing the decision-action cycle to fractions of a second, giving the decisive edge to the side with the more autonomous decision-action concurrency. At the operational level, commanders will be able to “sense,” “see,” and engage enemy formations far more quickly by applying machine-learning algorithms to collection and analysis of huge quantities of information and directing swarms of complex, autonomous systems to simultaneously attack the enemy throughout his operational depth.
At the strategic level, the commander supported by this capacity “sees” the strategic environment through sensors operating across the entire theater. The strategic commander’s capacity to ingest petabytes of information and conduct near-instantaneous analysis of information ranging from national technical means to tactical systems provides a qualitatively unsurpassed level of situational awareness and understanding heretofore unavailable to strategic commander.
AI-powered assistive technologies—such as intelligent assistants, advanced interactive visualizations, virtual reality technologies, and real-time displays projecting rapidly updated maps—will come together to enable this situational awareness. This level of strategic understanding generates the capacity for a speed in command and control and concurrent and subsequent actions that will consistently dominate—at a time and place of our choosing—because our superior concurrency will consistently overmatch the enemy’s capacity to respond.
All of this reawakens the perennial conversation about the nature and the character of war. If, indeed, we are poised at the edge of hyperwar, we must explore the changes necessary to adapt to this new conflict environment. It will require understanding the moral dimensions of these advances, educating a new generation of leaders, and developing the AI-powered analytical systems and autonomous weapons platforms. The mental, moral, and physical challenges of hyperwar demand analysis and a searching conversation. Our adversaries and our enemies are moving forward aggressively in this area. The United States must make the strategic investments both to be ready to wage hyperwar and to prevent us from being surprised by it.
Authors’ Note: As we build the conversation for hyperwar, we intend to recognize fellow travelers in this journey. To that end, we would like acknowledge the work of Peter Singer and August Cole in their excellent book, Ghost Fleet, an important fictional treatment of future war.
General Allen has served in a variety of command and staff positions in the Marine Corps and the Joint Force. He served as Special Presidential Envoy for the Global Coalition to Counter the Islamic State of Iraq and Levant, Commander of the NATO International Security Assistance Force in Afghanistan, and Deputy Commander of Central Command. He serves on the board of directors of several firms including SparkCognition. He was the co-recipient of the 2015 Eisenhower Award of the Business Executives for National Security.
Mr. Husain , recognized as Austin’s Top Technology Entrepreneur of the Year and one of Onalytica’s Top 100 global Artificial Intelligence influencers, is a serial entrepreneur and inventor with more than 50 filed patents. He is the founder and chief executive officer of SparkCognition, an award-winning machine-learning/AI-driven cognitive analytics company. He is the author of the upcoming book, The Sentient Machine, to be published by Simon & Schuster.