Federal agencies looking to transition from artificial intelligence (AI) research and development (R&D) to operations must overcome challenges both unique to their agencies, as well as governmentwide obstacles.
During GovConWire’s AI: Innovation in National Security Forum, Federal leaders zeroed in on how agencies can capitalize on the potential of AI technologies while still ensuring national security remains strong.
Dean Souleles, former chief technology advisor to the principal deputy director of national intelligence at the Office of the Director of National Intelligence, highlighted that both national security concerns and the government’s slow-moving acquisition process can cause delays in adopting AI technologies.
“The government is slow, and we have processes in place to allow for deliberate actions when making purchasing and acquisition decisions. That runs counter to the quick adoption of technology,” said Souleles.
The Department of Defense (DoD) also faces challenges with AI due to its unique operating environments and mission sets. The DoD operates in high-risk environments, and the cost of system failure or inoperability is high and potentially life-threatening. Warfighters need to know that they can depend on their systems to be resilient, secure, reliable, robust, accurate, and usable.
“The stakes in the DoD are incredibly high if we are to maintain our strategic edge. And there are multiple levels of uncertainty about how these systems might operate,” said Dr. Jane Pinelis, chief test and evaluation assessment at the Joint Artificial Intelligence Center. “Human uncertainty over whether the system is trustworthy and whether confidence in the system is justified. And AI system uncertainty has to do with whether the system itself can detect whether it’s operating outside of the environment it’s trained in or perhaps crossing some behavioral boundary.”
The transition to AI-ready systems within the DoD and national security space will require implementing methodical and highly deliberative processes for collecting and curating data. Dr. Jill Crisman, the principal director for AI at the DoD, is responsible for developing the department-wide AI roadmap and unifying and coordinating the department’s plans and investments to achieve a competitive advantage in AI.
A core concept of the roadmap is what Crisman calls the AI hubs. There are several AI hubs, one for each specific area of AI. And the goal, according to Crisman, is to create microservices that can apply to many end-user application systems.
“We plan on bringing the expertise and data across the department to work on common AI problems to increase efficiency and create better AI solutions that can integrate into systems within the DoD and our partners,” said Crisman.
Ultimately, according to keynote speaker, Katrina McFarland, commissioner at the National Security Commission on Artificial Intelligence, to effectively adopt and deploy AI solutions there needs to be proper leadership and a digitally savvy workforce, a clear understanding of the hardware, and innovative investments.