Artificial intelligence and machine learning are providing many efficiencies, but many government agencies are still in their early stages of implementing the technology. Leaders at the Defense Department are sharing how they have overcome challenges to see the technology making real impacts.
For one, it's a challenge getting people on board, said Naval Air Systems Command's Command Data Officer Michael Pomatto.
“Understanding the value of what [machine learning] really brings to help reduce some of the confusion and the barrier for entry for those junior-level data science resources. The cleaning of data is a major focal area in the command and trying to educate people in this area,” Pomatto said. “Many times, when we talk about AI, you talk about deep-learning models. but one of the challenges I see is, don't start with a deep-learning model — start by asking the right research question.”
AI is essential to addressing back-office tasks because it allows humans to do what humans do best, said Defense Innovation Unit's AI Program Manager Jaime Fitzgibbon.
“What can machines do really, really well to reduce the burden from humans to allow the humans to do what they do really, really well,” Fitzgibbon said at a FedInsider webinar. “Even just finding ways to increase readiness of ships or aircraft or weapons systems — those are phenomenal areas where there is an analog in the commercial space to come and map those solutions in, fairly rapidly and getting it deployed.”
Fitzgibbon also believes leveraging machine learning around health tech could help to solve massive problems across all services.
“Veterans Affairs has tons of medical data, and with ML you could tap into images and train a model to determine an early stage of a cancer. That is a way of leveraging the rich data set we have, protecting privacy and deploying things that will take care of the warfighter in a very real way," she said."
Implementing AI along with automation also saves time for the workforce to focus on more critical work.
“From a quality assurance perspective within our M365, people are using flank speed right now to do configuration management. That’s something that an E4 or O1 would do before manually, but now they’re automating it and that’s time saved that they could focus on a more gratifying side of their job and fine tune that skill,” said Navy Technical Director of PEO Digital Justin Fanelli.
Fanelli highlights how applying AI/ML along with automation has beneficial impacts to handling large amounts of data.
“Honing in on exactly what data to look at is something the Black Pearl team has taken initiative on and working with that same super Advana team we can translate that into value,” Fanelli said. “Because flank speed is up and running, we have said come and get it and make what you can of this. Someone from NAVAIR said we saved $3 million a year in manual tasks by automating this.”
For all three organizations, AI/ML is important to moving the needle for their missions.
“The more money we save on IT, the better we collaborate, the more we can increase retention and have the best people on the military and civilian side working within this very meaningful space," Fanelli said. "It leaves more money to buy ships and to do things that help with the fight."
“We have to start modernizing today, we have to keep our competitive edge both commercially, but also in the international security space. We have to identify risks before they become a catastrophe and a crisis. That is what all of this is all about, is that back office to help us fulfill that vision,” Fitzgibbon added.
Pomatto said NAVAIR is looking at the consolidation of multiple AI/ML initiatives involving automation, which leads to reuse, which leads to consistent results and improved quality. The command is also looking for ways it can use AI/ML to find vulnerabilities within its security network.
“Understanding how our AI/ML capabilities can be leveraged in order to reduce costs and increase aircraft availability is really crucial to us in maintaining that competitive edge over our adversaries,” Pomatto said.