Much has been written about how Federal agencies are moving “full steam ahead” on artificial intelligence (AI), integrating the technology into operations and complying with the Biden administration’s sweeping AI executive order (EO).
At the Department of Justice’s Executive Office for United States Attorneys (EOUSA), however, Alex Reber has a more basic focus: modernizing an IT system that has been running, in some instances, since the 1970s.
That means a full-scale integration of AI will have to wait, Reber told attendees at the Red Hat Government Symposium today. EOUSA provides executive assistance and supervision to the nation’s U.S. attorney’s offices.
“We have 94 different districts across the country, each with their own political appointees, each with their own set of judges who each have their own rules for how things need to be formatted,” said Reber, the operations manager for EOUSA’s Enterprise Application Division. “When we look into AI being able to take that information and give us meaningful results out of it, we need to clean up the underlying data, and that’s the next three to five years … we’re a long way from that.”
Reber added that he sees shorter-term possibilities for AI to help “troubleshoot issues with (our) environment” – and he reported progress with EOUSA’s overall modernization, saying the number of internet servers has declined from more than 200 to about 50. “We’re reducing by 75 percent the amount of infrastructure we have to maintain,” he said.
Reber appeared at a panel discussion on “Evolving Application Delivery and Secure Software Supply Chain in the Age of AI.” Speakers from other Federal agencies described aspects of their own modernization journeys, focusing on trends such as increased use of containerization with tools such as Kubernetes.
Daniel Rubenstein, infrastructure architect in the FBI’s Criminal Justice Information Division, said the bureau is far along in the move to the cloud it began about seven years ago.
“We are probably 95 percent in the cloud,” he said. “In our divisions, it’s all the fingerprint matching, the facial recognition, the National Crime Information Center. So, we’re really focusing on maturity, cloud maturity … and that developer onboarding experience.”
E.P. Mathew, deputy chief information officer at the Defense Intelligence Agency (DIA) – which provides intelligence for the secretary of defense, the Joint Chiefs of Staff, and combatant commands – said the agency has also zeroed in on bringing developers aboard.
But he said officials have faced challenges recruiting younger people for DIA and the intelligence community generally, an issue he saw with his own son, who is a software engineer. “He just came out of college … and he said, ‘Dad, [the intelligence community] doesn’t have access to a phone, or my Apple Watch, or Instagram, or all these other things. I can’t compete.’”
Mathew said DIA has tried to compensate by developing “an unclassified software factory” and letting applicants know that they don’t necessarily need high-level security clearances, which has helped with recruiting.
His words prompted session moderator Eamon McCormick, vice president for North America Public Sector Technical Sales at Red Hat, to joke: “You might have a chance at your son now.”
DIA has made its own progress in modernization, with officials revealing recently that the agency is nearing completion of a nearly two-year long “tech refresh” of its Joint Worldwide Intelligence Communication System (JWICS). The refresh is focused on updating equipment, building out cybersecurity tools, and optimizing use cases.
The FBI, meanwhile, has been increasing its use of AI “at a relatively basic level,” such as assisting with language translation and triaging data, FBI Deputy Director Paul Abbate said recently. He stressed that the bureau needs more technology resources and tools to better execute its mission.