Sixteen trillion dollars. That is what the artificial intelligence enterprise is expected to grow to by 2030, and the Joint Artificial Intelligence Center is taking steps to ensure that the Defense Department is at the forefront of this progress.
The benefits of AI include helping agencies solve complex problems, make key decisions and manage massive amounts of data.
One of the main goals of JAIC, when it was established in 2018, was to accelerate DOD’s adoption and integration of AI to achieve mission impact at scale.
In the beginning JAIC was focused on adoption and promotion of adoption, but now its attention is turning to the role of integration.
JAIC Director Lt. General Michael Groen said integrating these capabilities into a cohesive enterprise is going to require a purposeful vision.
“A purposeful vision for the platforms, a purposeful vision for how we’re going to use data, a purposeful vision for how we’re going to build a supporting architecture that supports a concept like JADC2,” he said during a recent virtual event.
According to Groen, culture is one of the biggest hurdles DOD is facing when it comes to AI and said the department needs to switch to a software mindset where things are built to continuously evolve.
“That sort of continuous capability development in a software environment is a real challenge culturally to the department because our processes are gates for hardware programs,” he said. "Everyone in the department is working to how do we do this faster, how do we recognize like what it means to be a software integration organization?”
Another cultural shift that DOD will have to deal with is the mindset of enterprise data.
Data is collected from people all over the place, but that data doesn’t belong to the person who collects it, it belongs to the enterprise.
“You’re collecting data so that the joint force can be more successful, whether its weather data, location data, target detection data, or the heart rate of a soldier in a squad because you’re monitoring your force down to that level that’s possible in this environment, but keeping all that data straight,” he said.
It is vital to catalog data and have a solid data enterprise. You should know what you have, where it is and how to use it.
“Then magic begins to happen because now you can start to doing analytics, now you can start doing AI and now you can start to really proliferate the tools that need data as a fuel, so you get your fuel source laid down, then you can start bringing vehicles to the fight,” he said.
In September, JAIC launched new data cards to help address the data challenges across DOD. Data cards, which are similar to library index cards used to control both data sets and algorithms, allow a user to identify data that might be useful to their decision-making and integrate it into their process.
“You can look up your data quickly to say, ‘Hey, I need data from this sensor or about this phenomenology on the battlefield,’ and you can quickly go find that data and then quickly operate on it because you understand the parameters of that,” he said.
Groen added that you could make some big mistakes if you didn’t have a system like data cards.
“For example, if the Army builds an AI algorithm that depends on a data stream coming out of the Air Force, well then we need to make sure as a joint force that the Air Force knows that there are these dependencies on their data across the force. So, the data card methodology helps us start to get organized in that way. So that we can identify dependencies where we can say whoa, whoa, whoa don’t stop that data because we are using that for these decisions over here, “ said Groen.
This year DOD also introduced the AI and Data Accelerator Initiative (AIDA). It’s an experimental program that works with combatant commands and helps build a decision environment for them.
AIDA is designed to help the department make rapid progress in the right direction.
“AIDA is going to make us fast through process not through phonetic activity, but by building capabilities and then archiving those capabilities and pulling those capabilities together,” he said.
The platform environment is another important component. Groen believes AIDA can help in making sure that the platform environment is intact.
“Data platform environment, a developmental platform, what we call our fabric, that stitches together our developmental platforms across the services and the JAIC and then we also want to make sure our operational environment, where those operational AIs live and work when the data is flowing,” said Groen.
AI will also be able to assist DOD in another crucial area: auditability.
“One of the things the department has pursued relentlessly is auditability. You can use artificial intelligence to help you on your journey to audit ability by identifying unmatched financial transactions, by identifying policy mismatches or process mismatches,” said Groen.
DOD is still trying to figure out more ways to incorporate AI, but there are many success stories across the services in contributing to the departments overall JADC2 concept.
“The Navy has Project Overmatch, the Air Force has [Advanced Battle Management System (AMBS)], so everybody kind of has like the idea,” he said. “The environment has shifted, and we have to recognize it.”