Please ensure Javascript is enabled for purposes of website accessibility

A smart defense

In late 2020, the Air Force made headlines when it announced that an artificial intelligence co-pilot, named ARTUµ, helped command and control a U.S. military spy plane for the first time in history.

If the name, pronounced R-2, sounds familiar, it is. Think R2-D2, or “Artoo,” Luke Skywalker’s lovable droid and X-Wing copilot from the “Star Wars” franchise. Except this was not a galaxy far away, but Beale Air Force Base in California.

ARTUµ controlled sensors and tactical navigation of a U-2 Dragon Lady on a reconnaissance training mission out of Beale on Dec. 15, 2020. It was charged with searching for enemy missile launchers while the plane’s human pilot, known only by the call sign “Vudu,” searched for enemy aircraft during a simulated missile strike.

The tech, developed by McLean-based Fortune 500 contractor Booz Allen Hamilton and Air Force researchers, modified an open-source gaming algorithm and ran more than 1 million training simulations in a lab — a “digital Dagobah,” Will Roper, who then served as the service’s assistant secretary for acquisition technology and logistics, wrote in an editorial for Popular Mechanics. ARTUµ was mission-ready in just over a month.

“Failing to realize AI’s full potential will mean ceding decision advantage to our adversaries,” Roper said at the time. 

Fast-forward four years and AI’s technological advancements have continued, transforming lives and — controversially — livelihoods as it becomes more entrenched in the workplace. At the same time, the military has continued to cite how critical AI will remain in helping the U.S. outpace its adversaries. Even as the Pentagon faces ongoing tensions in the Middle East amid the conflict between Israel and Hamas, and in Europe with Russia’s ongoing war on Ukraine, military leaders are refocusing for the potential of a wide-ranging battle with China in the Pacific that would most likely unfold across sea, air, land, space and cyberspace.

Virginia’s defense contractors are at the cusp of that work, with a hand in some of the largest and most transformative AI projects on behalf of the military. Those range from warfighting tools like unmanned vehicles to generative AI software to perform mundane business support tasks like military personnel record searches. That’s work that could be game changing for the military at a time when budgets and manpower are tight and harnessing data could be key to maintaining the upper hand against an adversary.

“We try to focus on mission meets
innovation,” says Holly Levanto, who is overseeing delivery of AI and digital solutions for Booz Allen Hamilton’s defense clients. Photo by Shannon Ayres

“I think that the eye is on the prize … when it comes to [the Department of Defense] right now, from the perspective of this is something we have to do from a national security point of view based on threats that we see from other nation states,” says Jason Payne, chief technology officer for Arlington County-based Microsoft Federal, which currently has a contract worth as much as $21.9 billion to produce more than 100,000 AI-enhanced goggles for the Army. “We know that near-peer competitors are investing heavily in this technology.”

Crunching data

The Pentagon is also investing heavily in AI technology. Its fiscal 2025 budget request, which totals $850 billion and was released in March, includes $1.8 billion for AI spending as well as an additional $1.4 billion for the department’s Combined Joint All-Domain Command and Control project, an ambitious departmentwide effort to connect “sensors to shooters to targets” globally.

But those dollar figures, the Pentagon admits, don’t likely tell the full story. With AI involved in so many programs, the Pentagon’s comptroller has acknowledged it’s difficult to provide a detailed breakdown of its AI investments. Even pinning down the exact number of AI defense projects is challenging. A 2022 Government Accountability Office report found that the DOD had at least 685 ongoing AI projects spanning the military service — a figure based on procurement and research and development dollars.

While those numbers may not offer a ton of clarity on the scope to which the Pentagon is looking toward AI, they do underscore the importance of it for the military, and Virginia contractors are benefiting from that desire.

Booz Allen Hamilton bills itself as the largest supplier of AI services to the federal government, with more than 300 active projects involving AI, according to Holly Levanto, a vice president overseeing the delivery of AI and digital solutions for Booz Allen’s U.S. defense clients.

“We try to focus on mission meets innovation,” Levanto says.

That work has included some of the Pentagon’s largest AI projects to date, including an $800 million, five-year task order awarded in 2020 to integrate and develop AI for the warfighter in the Alliant 2 Joint Warfighter Task Order, as well as a $885 million, five-year task order awarded in 2018 to help the DOD sift through its enormous amount of reconnaissance data — a project called Enterprise Machine Learning Analytics and Persistent Services, or eMAPS — through the deployment of AI and neural and deep neural networks. Booz Allen won a $1.5 billion recompete for the project in 2022.

Focusing on those mission areas has meant incorporating AI in ways to parse data faster. As an example, the Navy gathers vast amounts of data from its ships, Levanto says. Booz Allen has turned raw naval message traffic into tabular data that can be more easily and quickly analyzed to pinpoint trends. 

“We can send AI models to the edge at the point of data collection,” says Levanto, a former naval surface warfare officer. “And so, we have some real-world scenarios where we’ve done that in points on the battlefield.”

Booz Allen also launched a venture capital fund in 2022 to sharpen its tech capabilities. It has now invested in 10 companies, eight of which are AI-focused, Levanto says. That included an investment in Wisconsin-based RAIC Labs, which developed a model- generating platform using unstructured data. In 2023, RAIC made headlines when its tech was used to track a Chinese intelligence balloon that traveled over the U.S. before being shot down by a military jet off the coast of South Carolina.

“Our ultimate goal is to get the Department of Defense to be able to utilize these leading commercial technologies … and so we need to help bridge that,” Levanto says.

Falls Church-based General Dynamics Information Technology, a subsidiary of Reston-based Fortune Global 500 aerospace and defense contractor General Dynamics, is also no stranger to big defense contracts involving AI, or those that involve wrangling large sets of data.

In March, GDIT received a $922 million contract to modernize enterprise IT infrastructure for U.S. Central Command, which directs and enables U.S. and allied military operations across the Middle East and a portion of Africa.

Data is the biggest barrier to AI, says GDIT’s Brandon Bean, the AI and machine learning leader for the company’s defense division. That includes data quality and integrity as well as accessing old, siloed IT architectures. Where it used to be that applications were built to create data as a byproduct, the paradigm has shifted. Now, data “is what the application is built to support,” Bean says. “The data comes first.”

At a September conference hosted by GDIT at Amazon’s HQ2 headquarters, John Hale, chief of cloud services for the Defense Information Systems Agency, discussed how DOD is working with contractors to update antiquated computer code with AI.

“We’re using AI capabilities to … modernize legacy code that all the people who ever wrote it are long gone,” Hale said. “And you know, it’s not perfect, but it gets us like 85 to 90% of the way there, and then we’re able to manually fill in that last 10 to 15% to bring these applications into the 21st century.”

For CENTCOM, GDIT is tasked with creating data analytical services to support decision-making across nearly 20 networks and building data centricity and literacy across the command. By leveraging AI, including incorporating data tagging, what has previously required a more tedious process of manual data sampling of mountains of records can be extrapolated much faster, giving commanders the potential to better evaluate what worked during missions, or develop trainings based on lessons learned. It could also help service members to prove justifications for injuries that may not have been recorded in their medical records, Bean says.

AI in the cockpit

While AI is helping the Defense Department wrangle large amounts of data for higher level decision-making, the Pentagon is also incorporating AI in weapons systems and for operational use by warfighters. And that tech is getting increasingly advanced.

In May, Air Force Secretary Frank Kendall, who has advocated for the military’s use of AI, rode in an F-16 Fighting Falcon that was controlled by AI in a dogfight exercise against another F-16 flown by a human. Relying on sensors, California-based Shield AI developed the program used by the Air Force during the flight. In March, Arlington County-based Boeing announced a collaboration with Shield AI to develop autonomous and AI technologies for defense programs. Boeing declined to comment for this story.

With AI in the cockpit, the technology shows no signs of slowing down, including in a variety of unmanned vehicles, which will be a key component in future battles, with several drone initiatives underway by the Pentagon and military branches.

At Newport News-based Huntington Ingalls Industries, Virginia’s largest industrial employer and the nation’s only builder of nuclear-powered aircraft carriers, computer vision and recognition technologies have improved to the point where autonomous undersea vehicles like the company’s REMUS platform can be used to hunt for targets, gather intelligence and respond to findings without having to report back to the surface, says Andrew Howard, senior director of unmanned surface vehicles and autonomy programs within HII’s Mission Technologies division.

“Based on … customer comfort with things, different use cases, they could either update its survey pattern based on that information, or they could use that as the cue to pass information back to a surface operator to … take action based on that,” Howard says. “So, it’s really kind of made the information a bit more actionable than it used to be.”

The Navy in December 2023 announced that it had successfully launched and recovered a REMUS “Yellow Moray” drone via torpedo tubes on the USS Delaware, a Virginia-class attack submarine commissioned in 2022 and built by HII in partnership with General Dynamics’ Connecticut-based Electric Boat subsidiary. The Navy has said it could field the program for its submarine fleet later this year.

Meanwhile, the Marine Corps has been testing similar surface-level technology for its Long-Range Unmanned Surface Vessel (LRUSV) program using technology developed by HII that uses cameras and machine learning to identify and classify targets for maritime domain awareness, Howard adds. The drone’s tech passes intelligence to an operations center for action. Based on that feedback, the drone can then update its mission and shadow an intended target if called upon to act.

Gathering intelligence with less risk to warfighters can help save lives. Making that information more readily available can make work easier, too.

Reston-based Fortune 1000 contractor CACI International offers the DarkBlue Intelligence Suite, a tool that incorporates various AI techniques, including computer vision and image processing, to help analysts in dark web investigations and tracking. The company received a $239 million six-year task order in August to provide intelligence analysis and operations, including the DarkBlue suite, to the Army’s Europe and Africa command.

AI is also helping the Marines step into the metaverse. In October 2023, Fairfax’s CGI Federal, the U.S.-based arm of the Canadian professional services and consultancy, announced that it successfully completed a $34 million pilot to digitally twin the Florida-based Marine Corps Platform Integration Center’s assets into a virtual world by tagging its inventory and helping the service track its assets in real time. Being able to keep up with equipment like tanks as they travel the world could be of huge importance in a distributed battle across the Pacific, where troops could set up on airfields constructed on austere island chains. It could also help the service track maintenance needs and predict trends across vehicle fleets, says CGI Vice President Stephanie Ackman, who leads the company’s technology practice for defense, space and intelligence clients. 

“When the rubber meets the road … [does a taxpayer] care about where the stuff is?” Ackman asks. “Yes, but they care more so about the safety of the individuals that are down range.”