With Federal agencies beginning to implement AI across their operations, the Defense Advanced Research Projects Agency (DARPA) – the main research and development agency of the Department of Defense (DoD) – is looking to speed up its efforts to fold AI innovators and contractors into the agency’s research work.
Jason Preisser, director of the Mission Services Office at DARPA, discussed how the agency is increasing the velocity of its AI work during an April 8 webinar hosted by the National Academy of Public Administration (NAPA) and GovExec.
“The main thing that we’re focused on is increasing velocity through the process, from ideation to getting performers or contractors on contract and doing national security research,” said Preisser.
“The way we’re looking at AI from a business process automation and machine learning perspective, as well as leveraging large language models (LLMs) … is to where in that process … can we increase that velocity,” he said.
Part of that effort involves hosting events to spread the word, Preisser said. “We’re holding these kinds of workshops, with all the functional leads – whether it be contracting, comptrollers, security, they all have a role in getting those performers doing that national security research,” he said.
“As I think about AI, it’s simply [about] how can we leverage this new artificial intelligence tool that’s out there … to help increase that velocity to increase our productivity and move things faster through that chain,” he explained.
Preisser said that while the agency will move towards the goal of creating more AI use cases, he added that some of the data being used in business process innovation is still in its “infancy.”
“I think we’re still learning as to what kind of expertise we need and where we need to place that within our organization,” he said.
“DARPA is very small, it’s only about 240 government folks,” he continued. “That gives us an advantage that we can be very agile and nimble as to how we approach data. But I think that’s one thing we’re going to have to really analyze over time: what datasets we are going to need to use and how much curation is required to make that useful,” the official said.
