The logic of war is simple: get there fastest with the mostest. Yet, as Clausewitz reminds us, even if war sometimes is simple, the simplest thing is hard. Nothing is harder than anticipating the next war. As peacetime political processes demand predictability, especially in democracies, preparations are often biased toward fighting the last war. Our current predicament is that the distance between the last and the next can be too large to bridge if we do not catch up with the fundamental changes created in everyday life by digital, networked, and AI technologies.
Imagine fast and massive sieving through satellite pictures to identify targets in real-time while firing solutions can be fed to any nearby unit, Uber-like, ensuring targets are destroyed within minutes of discovery. Even better, through signal recognition, enemy radars and electronic warfare units become tin electronic ears by false signaling or spoofing. And this does not consider the scenario where they end up in a ball of wires and burned metal sheets.
These are not future scenarios; they are becoming reality every day. Yet, adopting AI technologies at scale trails behind. One reason is fear. Many believe that there are high chances that AI technologies will go awry, from targeting mistakes that no one will be responsible for the possibility, as remote as it might seem, that the AI-powered devices might turn against their handlers.
The U.S. research and defense world is caught between these two competing forces. Some, such as Eric Schmidt of Sun Microsystems and Google fame, believe that the time to invest in AI for defense was yesterday; others, including the U.S. State Department, are not waiting for workable warfighting solutions before entangling them in regulations.
If this discussion were a theoretical one, with no peer competitor breathing down America’s neck, the debate could be enlightening. However, the world has entered a period of active conflict. The Communist rulers of China and the Russian political elites supported by a number of free agents, such as Iran or North Korea, are furiously investing in new classes of AI-powered war technologies, from the most visible (drones) to the barely conceivable (space attack robots).
Yet, despite the clear and present danger represented by these looming threats, some voices insist not only that AI investments demand more careful analysis and vetting but that competitive concerns can be set aside. A recent article in the Foreign Affairs listed a litany of reasons why China might in fact be behind, from the lack of advanced hardware infrastructure to the heavy-handed approach of the Xi Jinping administration toward AI, which it can conceive as a potential internal security threat. These voices would dampen our enthusiasm for AI military technologies with a backwards approach: “regulate today” and “plan tomorrow.”
This belies the urgency of the situation. Even if the pacing competitor, Communist China, is capable of half the things it claims to be capable of, it would be strategic blindness to wait until the adversary will complete its plans when ours are in fact, not even starting to take off the ground. Example in point: China’s Multidomain Precision Warfare concept, which aims to weave all data, decision streams, and fire potential into one integrated firepower fist that can be brought to bear across domains (air, space, land or sea) or geographies. Developed since the mid-2000s, the system is expected to power all future operations, especially the impending military conflict around Taiwan.
On the other hand, the much-vaunted Joint All Domain Command and Control (JADC2) concept envisaged by the United States is in the proverbial “evolving phase.” Stymied by conceptual fuzziness, lack of doctrinal integration with the U.S. warfighting culture, and simple lack of understanding of what it can be, JADC2 lingers in a limbo of hopes tied up in bureaucratic tape. Instead of offering an immediate revolution in how targets are acquired and fire executed at the speed of thought, JADC2 resembles the stereotypical plot of the Sasquatch literature: “Was it or wasn’t it?” Just a few months ago, the Admiral in charge of the process, Susan BryarJoyner, spent an entire keynote address at a major conference to convince the audience that the concept is a verb, not a noun. (Is it?)
A national technology strategy moves with the momentum and cashing force of ocean liners. They are hard to start, hard to stop. Their crushing force is, however, incredible and, to a point, incomprehensible. The Manhattan Project was one such strategy: true war-stopping and, for a while, war-preventing. The AI revolution in military affairs needs its Manhattan moment.
Sorin Adam Matei, P.hD., is the associate dean of research, FORCES initiative director and a Krach Institute senior fellow at Purdue University. He leads the Purdue University effort to launch the Strategic Defense Technologies of the Interdisciplinary Engineering MS program, which combines data, computer and military sciences.