Agencies Must Upgrade Networks to Handle AI
AI also ties in with another federal technology trend: the evolution of zero-trust security environments, which can be enhanced by AI capabilities that disrupt attackers, says Microsoft Federal Security CTO Steve Faehl.
“Secure-by-design AI system architectures will be critical to ensuring the benefits of AI while mitigating the risks,” he says.
No matter the eventual purpose, agencies will be focusing on making their existing systems and networks AI-ready, says Garrett Berntsen, director for technology and national security for the National Security Council, speaking for himself and not his agency.
“As more and more agencies start to pilot and scale AI applications, IT leaders are going to be asked to provide the necessary compute, storage and software stack to deliver these applications,” he says. “Creating AI capabilities in support of the mission will drive IT requirements for more robust and interoperable systems.”
READ MORE: What are the other ways that artificial intelligence can benefit federal agencies?
Generative AI Poses Different Challenges
Generative AI, which can create new content rather than just performing routine tasks, also poses challenges and promise. “We have really only scratched the surface of what it can do to help us be more efficient,” says Leslie Beavers, the Department of Defense’s principal deputy CIO.
It may be able to help agencies in use cases ranging from customer service to fraud detection, says Jonathan Alboum, federal CTO and principal data strategist at ServiceNow. He sees government adopting “smaller and narrowly focused language models designed for specific tasks in specific industries.”
Space Force Chief Technology and Innovation Officer Lisa Costa says that adopting generative AI will cause “transformative disruption and demonstrate benefits through content creation faster than many leaders can imagine.”
This may require a slower adoption rate while agency leaders adapt their IT infrastructures to handle AI and address worker mistrust of the tool.
“We’ve got to manage data sets to protect government personnel privacy and preserve operations security during the employment and integration of generative AI tools,” she says.
Beavers agrees: “As we lean into using these capabilities, we need to keep concerns about information security and cybersecurity in mind.”
