Air Force Needs ‘Disruptive’ AI, Data Processing Technologies To Drive New ISR Plan

The Air Force is looking to industry to deliver advanced machine learning capabilities for its new intelligence, surveillance and reconnaissance plan, which focuses on growing its data collection efforts across drone sensors, an expanding fleet of satellites and publicly available information.

Lt. Gen. VeraLinn “Dash” Jamieson, deputy chief of staff for ISR, said the new ISR flight plan will see the service moving toward more rapid prototyping and developing operations with industry to deploy data processing algorithms and AI capabilities at a faster pace.

“Industry really knows a lot more about that technology than we do. So we went out and we talked to various icons in industry in how they approach this,” Jamieson told attendees at an Aug. 2 Mitchell Institute event. “We’re going to have annex to kick off the flight plan that describes our priorities and goals for using machine intelligence. The reason is that in the future we actually have to have decision advantage at the speed of relevance.”

The flight plan specifically calls for disruptive technologies including AI, neural networks, deep learning, human-machine teaming, according to Jamieson.

Officials said the new plan includes details on acquisition reform that would allow the Air Force more opportunities to work with software providers on instantaneous updates for the expanded ISR enterprise.

“Software acquisition has to be done at the speed of fielding,” Jamieson said.

Jamieson first announced the ISR plan last month, which aims to move away from manpower-intensive approach at processing the Air Force’s data toward a greater embrace of AI to match the growing capabilities of near-peer competitors such as China and Russia (Defense Daily, July 26).

Air Force senior leadership said they needed new capabilities that allowed ISR operators to process data more effectively in increasingly contested environments and move away from independent, single-intelligence platforms to a multi-intelligence enterprise.

“What we see is that the traditional approach to ISR has been losing its effectiveness over time. As we look forward and project on the trends. It’s going to continue to do so. We’re not going to wait for it to become completely ineffective. What this flight plan talks about are the directions to move in order to maintain effectiveness and be the leader in this competition and not fall behind,” said Kenneth Bray, the Air Force’s associate deputy chief of staff for ISR. “Machine intelligence is going to enable our humans, our sensors and our platforms. It is going to be the center of what we do.”

The ISR flight plan looks to expand the Air Force’s sensing grid with data sourcing from drone activity, satellite collection and sorting through publicly available information, according to Jamieson.

“We looked at, not just taking it to the next step and saying ‘what’s the next pod I’m going to put on an MQ-9,’ really wanting preserved decision space for those disruptive technologies to come in to enable us to look at things like autonomous swarming, manned-unmanned capabilities,”Jamieson said.

Air Force officials are looking at the impending satellite boom over the next two to four years to drop down the time for whole earth imaging from hours to minutes.

“When you reach hundreds of satellites, you break through a glass ceiling and get to capabilities that are very interesting out in space. When you reach thousands in space, now you get to something truly interesting. That’s what we want to bring and what we’re recommending in the flight plan,” Bray said.

Jamieson said the Air Force has already started working with industry on developing algorithms to make better use of publicly available information pulled from social media sites.

“Typically, the intelligence force, and I’ll just speak for the Air Force, we really have looked at [publicly available information] and maybe used it at endgame. We’re going to flip that on its head. And we’re going to use publicly available information as a foundational piece of our information,” Jamieson said. “When you have, literally, billions of people tweeting and giving out information, you actually can use that information. You can’t use it in a linear, analog capability. But if you have a data strategy and machine intelligence and you’re working with industry on application and algorithm development, the first thing you do is you create algorithms to identify pure data.”





More Stories You Might Like