- Just as artificial intelligence is transforming the way companies across the globe operate, next-generation intelligent systems and software will modernize and enhance America’s defense capabilities.
- With new threats looming, AI can bolster and add new dimensions to our war planning capabilities and enable military leaders to make better, quicker decisions.
- Humans, aided by AI, could command a new fighting force of thousands of low-cost, uncrewed vehicles representing an unprecedented show of firepower, while saving the lives of servicemembers.
- Intelligent software will also allow for a more rigorous, data-centric approach to military spending, reducing the waste long associated with defense procurement.
- These technologies, which the Pentagon is already adopting, will ultimately serve to strengthen human capabilities, not replace them.
Over the past year, the idea that AI will transform how we work and live has become ingrained in the zeitgeist. Analysts everywhere are quantifying AI’s expected impact on business and the economy, and reports that measure it in the trillions of dollars—from the likes of McKinsey, PwC and others—land on newspaper front pages with regularity.
Much less understood and publicized is AI’s expected impact on one sector of vital importance to all Americans: our military. Just as next-generation intelligent systems and software are expected to drive significant value for businesses, AI is poised to transform just about everything related to our national defense.
With their capacity to analyze vast amounts of data, do sophisticated modeling and predictions, and automate increasingly complicated activities, AI systems will drive dramatic improvements in how the military plans for war, how it executes operations, and what it buys and when. As Sen. Joe Manchin put it at a Senate Armed Services hearing earlier this year, AI “changes the game” of war altogether.
This isn’t a hypothetical future. Major US adversaries are already investing heavily in AI-enabled software and hardware, from unmanned intelligent combat systems and next generation drones to systems and sensors that enhance battlefield situational awareness. Determined to remain the world’s preeminent military power for the 21st century, the U.S. is also embracing these and other technologies. The US National Science and Technology Council first published its National Strategic Plan for Artificial Intelligence Research and Development in 2016, which it continues to update, most recently in 2023. Today, the US Department of Defense (DoD) is pursuing close to 700 AI projects and has requested $1.8 billion for AI research and development in 2024. And last year, a machine learning executive with experience at companies like Lyft, Dropbox, and LinkedIn was appointed to the new position of DoD’s Chief Digital and AI Officer.
Augmenting, but not replacing, humans
We’ll have to tread carefully and responsibly. AI’s sheer power renders it controversial – even more so in military settings. It is encouraging to see the DoD taking a proactive approach to establishing an approval framework and testing standards for the use of “autonomous and semi-autonomous functions in weapons systems.”
Ultimately, we’ll have to ensure that AI is developed so that it does not eliminate human decision making, but augments it. That’s not only responsible, but also effective. We can look to chess-playing AIs as a guide for how the technology can best strengthen our military capabilities. In 1997, IBM’s Deep Blue beat world champion Gary Kasparov. For many years after, the best games of chess came neither from humans alone, nor from AI, but rather from a combination of human and artificial intelligence, or “centaur AI.” Similarly, AI will not diminish the value of our highly trained troops or our military brass. The specifics of how they do their jobs may shift, but AI will only make their skills and judgment more vital and effective.
“We should think of AI as an extension of the human operator that can help us both make and execute our decisions.”
With the accelerated progress we’ve seen in AI recently, I’m more excited than ever about how the partnership between human and machine can improve our national defense. It will make our military capabilities not only smarter, nimbler, more resilient and more cost efficient, but less dangerous for American servicemembers. There’s even an argument that using AI to operate unmanned machinery could confer new military power to democracies, making it politically easier to go into battle without human casualties and harder for aggressors to invade and attempt to control territories (which still requires many humans).
Here are three distinct ways I see AI supercharging our national defense:
- An “affordable mass” of autonomous vehicles
The F-35 fighter jet. The V22 Osprey aircraft. The DDG-51 guided missile destroyer. For decades, our military has excelled at building really big, highly advanced, and exorbitantly complicated vehicles and machines. In life and death situations, these systems bring state-of-the-art capabilities and are “survivable,” which means they’re designed to keep performing their mission and protect the service members onboard. But these impressive vehicles can take decades to develop, with the DoD only able to afford a limited number of them. To justify the high costs, they have to be operational for decades, at which point they may no longer be very useful.
A new approach called “affordable mass” is taking root. For example, the DoD’s “Replicator initiative” aims to modernize America’s arsenal by leveraging AI-enabled autonomous technologies. Instead of buying a few really big things, the military is looking at acquiring thousands of low-cost uncrewed vehicles that carry out human-directed tasks coordinated by AI systems. AI systems could execute human-directed tasks with greater precision and finesse than humans could achieve alone.
In a fight, not all of these vehicles will survive but the operation of the overall system is extremely robust. The Army, for instance, is testing driverless vehicles for use in high-risk convoys. This will mean fewer soldiers in a formation and fewer lives in harm’s way.
Similarly, the Navy has said it will spend an estimated $4.3 billion acquiring nearly two dozen uncrewed vehicles over the next five years, such as robot ships and submersibles operated by software instead of sailors. While the initial price tag may seem high, it represents only 3% to 4% of the Navy’s expected shipbuilding budget over that same period and, at a cost of about $200 million per vehicle, the unmanned submersibles are an order of magnitude cheaper than the $2 billion price tag of a Virginia-class submarine.
For its part, the Air Force is developing “loyal wingman” planes. Up to five of these next-generation drones can be paired to one human-piloted aircraft, multiplying the effect a single fighter pilot can have in battle.
“AI systems will drive dramatic improvements in how the military plans for war, how it executes operations, and what it buys and when”
Beyond these examples, I believe there is even greater promise in the kind of low-cost, unmanned aerial vehicles Swarm Aero and others are building. Instead of asking over-tasked pilots to supervise five other planes while in the air, I see value in military personnel away from the frontline managing large fleets of autonomous aircraft. Think thousands of small weapons-carrying aircraft swarming the sky and hundreds of submersibles churning the depth of the sea, all of them equipped with sensors that provide a holistic picture of the battlefield. The ability of these AI-powered swarms to collect and analyze granular, real-time information would help minimize the kind of intelligence failures that have caused existing weapons systems to hit the wrong targets, causing the tragic deaths of civilians.
In addition to saving the lives of men and women in uniform, an affordable mass of AI-operated autonomous vehicles can get us to risky places we can’t go otherwise and can represent an unprecedented show of force in areas critical to our current and future military defense. Our existing ships and aircraft carriers in the South China Sea, for instance, are within range of China’s long-range missiles. Sending them closer only makes them more vulnerable. But swarms of thousands of long-range unmanned vehicles can introduce the “fog of war” to the aerial battlefield, overwhelming detection and air defense systems and allowing our most valuable assets to enter the fray. Representing an overwhelming level of force that would saturate China’s air and sea defenses, UAV swarms could be an effective deterrent to aggression in the region.
- Smarter war planning
War games are a time-honored tradition. For over a millennium, they have helped military personnel explore tactical, operational, and strategic aspects of war, test their decision making, learn about new scenarios, and even formulate strategy. As war itself has grown in scale and sophistication, so too has our war gaming matured from table tops to computer models capable of simulating millions of variations on a single scenario. With the intricacies of modern warfare and the trend in affordable mass adding to an already large number of assets, even the smartest and most experienced military strategists supported by today’s computing capabilities can’t possibly assimilate and analyze all the bits of potentially relevant data available – from the performance of many types of sensors, radios, and jammers to information on specific weapons and vehicle capabilities. But software can, modeling legions of real-life scenarios and then identifying those most likely to succeed in a given context. That allows those in the war room to home in on the nuance of and compare only the most valuable of the myriad available strategies. Then, when a crisis hits, military leaders will be prepared with detailed and optimized responses more resilient and dynamic than either human or AI could design alone.
AI modeling can also help make us smarter during the chaos of conflict. Military strategists currently use data such as aircraft speeds and sensor capabilities to do tactical operational modeling and simulations of dynamic conflict situations. AI can take these simulated storyboards to the next level, with situation-specific tactical plans designed, stress-tested, and optimized in the advanced simulation environment for full impact.
While the ultimate decisions about what strategies to pursue will still reside with humans–people are much better at reacting appropriately to fundamentally new situations–AI systems can be designed with semi-autonomy so that they can sustain operations, like tracking an enemy asset, and react to new situations according to pre-programmed commands even in communication denied environments. We should think of AI as an extension of the human operator that can help us both make and execute our decisions. Machines equipped with AI are also highly reliable allies in carrying out human instructions, especially those involving large amounts of coordination.
- Smarter procurement
Building ships, fighter jets, and weapons systems to support what is the world’s most powerful military is an expensive, massively complex, multi-year process. And as the threat landscape has changed over the years, America’s force structure has often found itself struggling to adapt quickly enough. At times, weapons systems that seemed futuristic when they were ordered became obsolete, or failed to perform as expected.
The Pentagon has sought to fight the resulting inefficient allocation of resources with organizations like CAPE, or Cost Assessment and Program Evaluation, which uses data and analysis to help guide spending choices. AI would put these kinds of assessments on steroids, helping to create a robust internal system that continuously assesses the true value of all of its spending programs. The added muscle provided by AI would allow CAPE’s assessment framework to extend to the entire procurement process and uniformly across all branches of the military.
This system would evaluate not just individual programs, but assess how each one interacts with others to contribute to our strategic defense force — at the lowest possible cost. Is this ship or submarine addressing the current threat in a cost-effective manner? Based on what we know from AI-enabled war games and tactical modeling simulations, what assets will move the needle most in a particular fight? Such data-driven evaluations can also identify gaps in our future resources, helping to shape our understanding of what needs to be built. It can also help to redirect available dollars to the most efficient programs as well as to the clearest and most urgent needs. The Replicator program is designed exactly this way: “[W]e will not be asking for new money in [fiscal 2024]. Not all problems need new money; we are problem-solvers, and we intend to self-solve,” Deputy Defense Secretary Kathleen Hicks, the leader of the initiative, said recently.
“An affordable mass of AI-operated autonomous vehicles can get us to risky places we can’t go otherwise”
As the DoD prepares our national defense for the machine learning age, it will be essential to have rules about what AI can and can’t execute. The Autonomous Weapon System Working Group established at the start of this year will work to ensure new AI-enhanced armaments meet the military’s ethical standards and that those standards continue to evolve. These standards mean AI systems cannot be a black box. Developers of these programs will need to make them controllable and explainable, with reasoning behind how they “think” laid out in terms human users can understand. Instituting these guardrails will be critical, not only to avert a dystopian future of robots running wars, but also to ensure everyone, from policymakers to the public and our armed forces, can trust these intelligent weapons systems. They must align with America’s priorities and values as well as military objectives.
With the proper rules in place, AI is poised to be a transformative force that will strengthen America’s national defense. It will give our military new weapons systems and capabilities, smarter ways to plan for increasingly complex conflicts, and better ways to decide what to build and buy, and when. Along the way, it will help save both taxpayer dollars and, more importantly, lives.