INDUSTRY PERSPECTIVE AI’s Power to Transform Command and Control BY WES HAGA AND COURTNEY CROSBY By securely connecting sensors, data, decision-makers and weapons across multiple domains, the Air Force’s Advanced Battle Management System will move warfighters from radios and classified chat capabilities to the ubiquitous connectivity and availability that has long been established in the civilian world. It also offers powerful capabilities for com-mand and control. Artificial intelligence applications across the systems infra-structure will allow the service to take advantage of today’s sensor proliferation and robust communications paths to synthesize and harmonize data into actionable information at machine speed. This improves command-and-control posture — and, crucially, creates strategic advantages — by enabling new capabilities in information detection, identification and synthesis across multiple domains. How does this look in practice? Take for example anomalous traffic on a network, like a military patrol vessel sending sig-nals associated with civilian freighters. A human analyst might believe based on traditional intelligence that a civilian freighter was using the route. In this case, the analyst would fail to view the patrol vessel’s signals as a threat. An AI system, however, would detect underlying traffic pat-terns. It would more accurately flag the vessel’s signals as a cue to assemble nearby sea and air assets. In this scenario, AI saw something a human might not have caught and enabled infor-mation across domains to respond to the threat faster. There are more examples of how AI can support command and control at the strategic, tactical and operational levels, enabling commanders to more effectively direct forces and achieve desired actions such as faster and more accurate deci-sion-making. Warfighters and command-ers live by the OODA loop: observe, orient, decide, act. AI shortens and sharpens this deci-sion loop at every step. The first step, observation, is perhaps the most familiar: how AI processes vast amounts of information faster and more accurately. A human geospatial analyst scanning images for a specific black truck is limited to observing a single screen. The analyst may take several hours to find that vehicle — and may spot a navy blue SUV by mistake. By contrast, an AI-enabled computer vision system can scan dozens of feeds simultane-ously and more accurately identify discrete visual spectrums that the human eye can’t detect. AI also sharpens the orientation stage of decision-making by exceeding human cognitive limitations and processing speed. For example, an algorithm can train on millions of images in a single day; whereas a human analyst would be trained through months of coursework and likely still not accumulate the same number of images. As a result, an algorithm can generalize and process information that it has never seen before more effi-ciently than a human. Adding to its decision-making power, AI can predict the outcomes of a potential course of action and evaluate environ-mental or adversarial variables that would be beyond human capacity to assess. AI offers a capacity for course of action adju-dication and dynamic reprioritization for a broader spectrum of risk while “red teaming” courses of action in real time. At this critical decision step, commanders need to act in a timely fashion with precision and accuracy. Humans are lim-ited in the time they have to read reports, watch data feeds and make connections between disparate inputs and outputs. An AI system, by contrast, can ingest, process and synthesize vastly more information at superhuman speed. This empowers decision-makers with a fuller view of the “ground truth” when they need it. AI also strengthens the coordination of warfighting functions. Through its prediction capabilities, the technology enables warfighters to coordinate functions in new and innovative ways. For example, AI can identify and continue tracking a high-value target even when the target disappears from line of sight — like into a tunnel or behind buildings in a dense urban ter-rain. By computing using position, heading, speed of travel and similar factors, AI can not only maintain a track but also predict where the target will be 20 minutes from now to proactively mobilize assets. Now imagine there are two drones in proximity to the target, two and two-and-a-half miles away, respectively. While distance to the target may seem like the primary factor to consider in determining which drone should be employed, AI can calculate a large range of additional factors using predictive maintenance. By using real time data ingestion to understand the fuel gauge, range, time on station and other relevant data points, it could be revealed that the more distant drone will make it to the tar-get faster. In situations like this, AI offers huge potential for the dynamic retasking of assets. AI can also see ahead of the battlespace. It gives commanders and warfighters the ability to see forward and respond proac-tively rather than reactively. For example, an analyst receives a pattern-of-life report that indicates an adversarial military asset, like an airplane, has moved from its normal position to a bordering airfield. Human intelligence on the ground confirms this new activity. AI has the potential to connect these tradi-tionally siloed reports and scramble defensive assets while the enemy plane is still on enemy soil, rather than waiting until it is noticed entering friendly airspace. AI empowers connections between disparate pieces of infor-mation. By coalescing data across domains and services into a single place, it makes information more discoverable, for swifter and sharper decisions. It also allows learning and the application of knowledge across domains. It is particularly well-suited to addressing novel scenarios and changing mission requirements. Artificial intelligence gives commanders and warfighters a cursory understanding of the environment beyond current capabilities. It filters through vast sums of data to “tip and cue” anomalies. As AI systems process data and “learn,” they then “ AI EMPOWERS CONNECTIONS BETWEEN DISPARATE PIECES OF INFORMATION. ” 40 N AT I O N A L D E F E N S E • N O V E M B E R 2 0 2 0