‘Data centricity’ becoming increasingly important for DOD
The U.S. Department of Defense (DOD) continues to advance its efforts to centralize data. As military organizations continue to leverage data in operations, they will need rugged military tech that provides reliable access to that data.
Using data to better respond to the rapid pace of the modern military environment is increasingly key for missions, said Bill Streilein, chief technology officer of the Defense Department’s Chief Digital and Artificial Intelligence Office.
Speaking at a Defense One event Tuesday, Streilein emphasized the importance of “data centricity” for providing, synthesizing and analyzing information in real time to inform decision making.
“If everyone has access to the same data that is relevant they are empowered to make decisions much more quickly,” Streilein said.
By data-centered, Streilein means making data a primary and permanent asset in supporting missions. Data centricity is at the core of DOD’s plan for Joint All Domain Command and Control, which would connect all branches of the military through an AI-powered unified network.
In fiscal year 2024, the DOD is asking for $1.4 billion to further develop JADC2, delivering information on demand “across all domains and partners.”
Each of the services has its own JADC2 initiatives – Project Convergence for the Army; Project Overmatch for the Navy and Marine Corps, and the Advanced Battle Management System for the Air Force and Space Force.
Streilein said DOD is considering “data mesh technology” that would keep data at its source and would be accessible through a self-service catalog that data product managers would feed to support missions.
Streilein foresees a new role in IT for data product managers who work near the “data defense” sources to feed the network with information. “The manager is the person who understands how the data is extracted and connected with other data sources,” Streilein said.
When asked about other advances that AI will bring within the next five years, Streilein identified gains in “large language models.” The software understands, processes and produces human language by using massive amounts of text and data.
In the early stages, Streilein said large language models today are often unreliable and prone “to producing incorrect information,” as seen with ChatGPT. “The more we can assure the outcome in the future the better we can leverage them.”
“Large language models represent a real shift in how humans interact with AI,” Streilein said. “They present a natural interface. DOD may harness these to great effect.”
Overall, Streilein emphasized the importance of DOD connecting with academia and the commercial sector to identify and prototype emerging IT capabilities that have defense applications.
“Better access would enable us to reach out to smaller players where much of the innovation is happening,” he said.