he Air Force Distributed Common Ground System (AF DCGS), also referred to as the AN/GSQ-272 SENTINEL weapon system, is the Air Force’s primary intelligence, surveillance and reconnaissance (ISR) collection, processing, exploitation, analysis and dissemination (CPAD) system. (U.S. Air Force photo)

The Air Force Distributed Common Ground System (AF DCGS) is the Air Force’s primary ISR collection, processing, exploitation, analysis and dissemination (CPAD) system.

WASHINGTON: The Air Force is seeing limited success using AI “decision aids” to help operators sort through and integrate vast amounts of data to create actionable information for commanders, says Air Force Research Laboratory’s Jean-Charles Ledé.

But while the service has installed artificial intelligence into a handful of systems to do certain data-intensive tasks, it hasn’t figured out how to scale up AI to really make it meaningful across the service, he told the Potomac Officers Club 3rd Annual Artificial Intelligence Summit today.

“We don’t have a way of really going from these one-by-one, quick wins to a more general environment, where we can generalize this and make it available everywhere,” explained Ledé, autonomy technology advisor to AFRL’s Commander Brig. Gen. Heather Pringle. “And that’s the big issue that really needs to be addressed. … We need to move away from these single point demos to true scalability.”

AFRL has been a leader in developing AI for various service missions, including pioneering a unique industry collaboration to speed solutions to the toughest AI technical challenges — with one of its biggest efforts the DARPA AlphaDogfight contest. The collaborative, called the Autonomy Research Collaborative Network, or ARCNet, is a partnership between AFRL and an industry consortium managed by the non-profit SP Global Institute (SPGI).

Ledé said that “in general” the Air Force has been concentrating its initial AI initiatives around creating “decision aids,” as well as efforts to integrate AI into training — although the end goal is to push its use into all aspects.

“We look really at AI as potentially impacting everything we do in the Air Force, from the back office activities — things like, for example, preparing a response to a Freedom of Information Act, helping to edit those documents — to logistics, certainly; and eventually even combat application in either manned or unmanned platforms,” he elaborated. “So, the scope of application of AI is basically everything we do. And we’re starting to go down this path, but there’s still a lot of work to do.”

Transitioning AI demos into programs of record and integrating AI usage into all service activities is something the Air Force hopes the Joint Artificial Intelligence Center (JAIC) can help with,” he said. Tools such as the  Joint Common Foundation, a cloud-based platform that enables users to access Defense Department data and develop AI solutions in a secure environment, are aimed precisely at the question of scalability, he added.

One current Air Force success story, Ledé said, has been an AI system installed into the Distributed Common Ground System (DCGS) that connects most of the service’s airborne intelligence, surveillance and reconnaissance platforms — collecting data from the U-2, RQ-4 Global Hawk, MQ-1 Predator, MQ-9 Reaper and other platforms.

Also know as the AN/GSQ-272 SENTINEL, the venerable DCGS is used for ISR “planning and direction, collection, processing and exploitation, analysis and dissemination,” according to the service website. Former Air Force acquisition czar Will Roper revealed that DGCS was using an AI this past June but provided no details.

Ledé said today that the system has “increased by a factor of 10 the data production from analysts,” adding that a single “cell” of analysts “produced in three months more than the whole rest of the whole Air Force in a year when AI was deployed in that environment.”

Making better and faster use of imagery was, of course, one of the first projects undertaken as part of former DepSecDef Bob Work’s Third Offset. Known as Project Maven, it became a flashpoint for some Google employees, who protested the company working with the military. Project Maven is now a centerpiece of JAIC’s AI development experiments, and is being used by the Air Force in its Advanced Battle Management System (ABMS) program designed to build the technical scaffolding for Joint All Domain Command and Control (JADC2).

The Air Force awarded Raytheon Technologies a $178 million contract on March 10 for five years work to upgrade the legacy DGCS system of systems. It was a patching together of disparate, platform-specific data processing software. According to the service’s contract announcement, Raytheon will be providing “transitional mission support from legacy to open architecture infrastructure as the system completes open architecture modernization.”