With the resurgence of great power competition has come the resurgence of defense technology. Conflict with a near-peer competitor would look nothing like our recent warfighting experiences against nonstate actors. It could be fought over vast distances, involve technology we are unprepared to deal with and occur at a scale the United States has not had to contend with in decades. Suddenly, our glittering collection of large, exquisite systems, built to engineering standards decades behind consumer technology, feels outmoded.
Artificial intelligence should be the linchpin of our efforts to re-armor ourselves for a new kind of fight. In the near term, AI can help us to understand the world with greater speed and clarity than the human mind can achieve. The value of distributed sensors on manned and unmanned systems alike has long been apparent to the United States, but we face a manpower constraint that enables us to analyze only a fraction of the data we capture.
AI is capable — not will be, but is — of taking on the most time-consuming and mundane aspects of this analysis, simultaneously fusing information from thousands of data streams to present analysts with a complete view of what we know. With the cognitive burden of sifting through this information shifted to a machine, analysts could spend their energy determining how to act upon this information and where our remaining gaps lie.
With AI-enabled situational awareness, the United States military would be substantially better positioned to manage large-scale conflict with its competitors. But while awareness is crucial, AI can also enable newfound ways of defeating opponents who are hoping we fight as we always have. China’s buildup of “assassin’s mace” weapons should shock us into innovation: over-the-horizon radars and “carrier killer” missiles to detect and destroy our aircraft carriers; stockpiles of medium- and long-range ballistic missiles to overwhelm our forward-operating bases; and cyberweapons, anti-satellite missiles and other electronic attack capabilities designed to knock out our satellite communications.
Doubling down on our existing systems would be playing into China’s hands.
We have long recognized the value of shifting toward unmanned systems, and have indeed led the world in the development of exquisite unmanned platforms. AI, however, is the missing ingredient that can bring such systems to breathtaking scale.
Our current unmanned systems are, in truth, unmanned in name only, often requiring large crews of remote pilots and operators. Artificially intelligent systems, by contrast, can unlock an entirely new mode of distributed, flexible fighting at scale — or what the Defense Advanced Research Projects Agency calls “mosaic warfare.” With a radically reduced manpower constraint, the United States could discard its exquisite systems in favor of swarms of cheaper, artificially intelligent alternatives, which could conduct a range of missions with far less human oversight. Given that our competitors could target our brittle satellite communications systems, AI-enabled systems could and should be programmed to cycle through a variety of communications methods depending on what is available, including secure mesh networks, and should be capable of processing and sharing data on the edge without relying upon a central communications hub.
The potential value of inexpensive, distributed systems was on full display during 2020′s fight over Nagorno-Karabakh. Both Armenia and Azerbaijan knew their unresolved conflict would recur sooner or later, but only one side spent the interbellum years investing in attritable unmanned systems. Armenia’s expensive Russian anti-aircraft systems were dominated by relatively rudimentary UAVs, which targeted and struck them with near impunity.
Despite China’s and Russia’s public rumbling about AI, the United States has a number of advantages over its rivals in the race to develop sophisticated autonomous systems. The seven most valuable technology companies in the world are all American, as are three of the world’s five best computer science departments. We have the human capital, the wealth and the scale to develop world-class AI-enabled military technology that will help us make sense of a vaster and more complex battlefield, and act upon it.
Of course, we must turn these theoretical advantages into concrete reality. Encouraging our brightest software engineers to turn their efforts toward building AI for our national security means understanding how the leading technology companies develop software. Some of my colleagues have written about how the Department of Defense can shape its policies and practices to bring our best and brightest into the defense industry and grow the pool of technologically sophisticated nontraditional defense companies. The DoD, to its credit, seems to be listening. But we do not have time to pat ourselves on the back for speeches and articles about innovation. AI-enabled unmanned systems are available immediately, and that is when the DoD should be demanding them.
Brian Schimpf is the chief executive and co-founder of Anduril Industries.
