Positioning high-quality data as the foundation of the Department of Defense’s efforts to expand artificial intelligence technologies across its missions will require significant partnerships with the private sector, according to the Department of Defense Chief Digital and Artificial Intelligence Office (CDAO). the director said on Tuesday.
In a keynote address at CDAO’s Defense Data and AI Symposium, Chief Digital and Artificial Intelligence Officer Craig Martell said the agency is committed to providing tools for “the fight tonight” and providing interoperable technologies, among other things. He said he is trying to strike a delicate balance with long-term efforts to implement the system. A comprehensive automation solution across the DOD.
Regarding the development of AI solutions, Martell said the Department of Defense should “pay industry to build the models because they are always at the cutting edge.” But it added that the department is pursuing an “AI scaffolding” approach that creates “everything to the left and right of the model.”
Martel said the Pentagon only needs to “commission stovepipe solutions” that don’t use high-quality data to address some of its immediate needs, but that the department’s long-term AI implementation “is “Data quality is the most important aspect.” This is especially important when streamlining solutions across different combatant commands.
The Department of Defense’s Data, Analytics, and AI Adoption Strategy, developed by CDAO and published last November, positions data quality as the foundation of the department’s comprehensive “Hierarchy of AI Needs.”
The Department of Defense remains committed to fully implementing the Consolidated All-Domain Command and Control (CJADC2) initiative, which focuses on promoting interoperability between disparate military domains. Martel said the proper use of high-quality data (which he defines as “discoverable, accessible, and usable” data) will help make this happen.
“Imagine a world where combatant commanders don’t have access to information via PowerPoint or email, and the time it takes to gain situational awareness of what’s going on is reduced from a day or two to 10 minutes.” said Martel. , adding that by having that high-quality data available, combatant commanders can, for example, “build apps based on that data very quickly.”
But Martel said that to make this future possible, the Department of Defense must drive changes across the Department, especially when it comes to ensuring that quality data is used appropriately as part of the Department’s data adoption. “We need to rely more on the private sector to do this,” he said. – Mesh feature. Its purpose is to allow information to be shared between domains.
Martell said his interpretation of the data mesh is that “data stays where it is,” adding that “the best place for data is with the owner of the data.”
Martel also noted that CDAO had previously issued a request for information (RFI) last year seeking input from the industry on data mesh capabilities to support CJADC2 efforts. It is extremely valuable to have this support.” . He said more RFIs are expected in 2024, including those related to model monitoring.
“We want vendors to understand the rules and come to play, and model builders in departments to come to us and say, ‘We need to build the next kind of model.’ , we want to build this AI scaffolding,” Martell said. He said. “And we have a handbook with a list of vendors who know how to follow the rules or who are following the correct rules. And we’re working hard on this.”
Martel said the CDAO is also working this year to open government data and unified data catalog application programming interfaces (or APIs) to developers, creating opportunities for developers to compete with the Department of Defense. he said. Users can get direct feedback on their products.
