(Updates throughout with Government Accountability Office report information)
Bloomberg Government subscribers get the stories like this first. Act now and gain unlimited access to everything you need to grow your opportunities. Learn more.
The Defense Department’s office responsible for artificial intelligence integration wants to use its buying power to speed development of technology that can be used across military branches, and the federal government’s watchdog agency is recommending changes to make that happen.
The Government Accountability Office found the Pentagon lacks specific plans and timelines for buying artificial intelligence capabilities despite it being a top focus for the department’s modernization efforts, according to a report released Thursday.
Future military conflicts will require gear and software that haven’t been developed yet, and leadership of the Pentagon’s Chief Digital and Artificial Intelligence Office is thinking about how to make sure DOD is ready. It’s doing that by working with the private sector to identify a problem, find a solution, share data, implement feedback, and then deploy it “within days,” the office’s deputy Margie Palmieri said in an interview Wednesday.
AI and data are key elements of the Combined Joint All-Domain Command and Control, or CJADC2, initiative for enhanced decision-making. CDAO is focused on data integration, which includes making sure data is accessible and can be translated across the military branches and joint staff at the Pentagon.
Rapid deployment of new data analytics and AI tools is a primary concern for the Chief Digital and Artificial Intelligence Office, but common Pentagon procurement issues, including slow acquisition speed and adoption, also arise with AI-oriented technology.
Data and AI tools largely come from the private sector, and the solution for speeding up the process is in “how we write contracts” and how the federal government makes “data a product as opposed to making our systems the product,” Palmieri said.
The National Security Commission on AI reported to GAO that technology industry officials called the current federal contracting process “not well-suited to AI procurement and deployment.”
Palmieri said CDAO is working to find ways to change policies that delay or complicate the contracting process.
“Never have I seen an organization in DOD so empowered with the levers of change necessary all in one place,” she said.
The Defense Department has requested more funding for relevant resources, including $1.1 billion for core AI in fiscal 2023. The GAO in the new report said “establishing guidance would better position DOD to effectively spend funds on AI acquisitions consistently across the department.”
CDAO has made some of its AI acquisition resources available to various DOD components, and a project to catalog of those resources is underway by the Office of the Under Secretary of Defense for Acquisition and Sustainment, according to the GAO report.
The Intellectual Property Cadre, Defense Acquisition University, and the Air Force are planning to publish AI acquisition resources regarding intellectual property concerns, commercial AI training, as well as building and design of AI applications.
AI vendors concerned about protecting their intellectual property from competitors need to adapt their practices to meet certain DOD software requirements, which may be a disincentive to partner with DOD, according to the Intellectual Property Cadre. To encourage working with DOD, procurement officials should develop acquisition strategies that align requirements with AI vendor business models like using “other transaction” agreements for more flexibility to negotiate license agreements, the IP Cadre said.
CDAO’s TryAI is one path encouraging interaction between a potential vendor and DOD at no cost to the business and with no transfer of proprietary information to the government, Palmieri said.
“We can actually have a conversation on whether or not we should move forward to something and we don’t have to go through the RFP process or acquisition system,” she said, identifying the length of time standard requests for proposals can take to result in a new tool being adopted by a DOD office. “We can just put the capability on the table and figure out, ‘Is this the right thing?’”
TryAI is set up as a commercial solutions opening, one type of non-traditional procurement mechanism, and Palmieri said that for DOD to achieve its goals “we’re probably not going to do it with the normal contracts that we have.”
To contact the reporter on this story: Patty Nieberg in Washington at email@example.com
To contact the editors responsible for this story: Amanda H. Allen at firstname.lastname@example.org; Stephanie Gleason at email@example.com; Angela Greiling Keane at firstname.lastname@example.org