DHS to lean further into AI with generative pilots
Eric Hysen, the agency’s chief information officer and first chief AI officer, said the goal is to give employees access to new tools and technologies to deal...
HERSHEY, PA. — The Homeland Security Department will continue its aggressive lean into artificial intelligence.
In the coming weeks, DHS is expected to detail its plan for several generative AI pilots focused on the mission areas.
Eric Hysen, the agency’s chief information officer and first chief AI officer, said he’s spending about half of his time on AI issues with a focus on a how best to use and accelerate the technology to support mission areas.
“We view that we need to give our employees the tools they need to do their jobs. Our workforce is incredibly talented and incredibly passionate, and the workload that we have across DHS only gets larger every single day so when you look over the last several years, the number of new challenges that our workforce has taken on is truly incredible. So when we see a technology that has the potential to make our workforce more productive, we’re going to seize that. But we’re going to do that in a deliberate and appropriate way,” Hysen said after his panel at the ImaginationNation ELC 2023 conference sponsored by ACT-IAC. “We’re still finalizing our exact policies, but they will look like what is encouraged in President [Joe] Biden’s executive order that was recently signed, that talks about the importance of training for our workforce on how to use AI systems, that talks about policies and procedures around what information you can and can’t put into these systems, and then ensuring appropriate human review of those outputs.”
Hysen said declined to comment further on the pilots citing the need to finalize some details.
Biden issued the much-anticipated AI executive order on Monday, giving DHS several areas to focus on including establishing and leading an AI Safety and Security Advisory Board (AISSB) to support the responsible development of AI and assessing potential risks related to the use of AI in critical infrastructure sectors and around cybersecurity.
Hysen said the AI Safety and Security Advisory Board is modeled after the Cyber Safety Review Board, created in February 2022 to examine significant hacking incidents and recommend improvements.
“It also involves ensuring that AI cannot be used to generate threat information on weapons of mass destruction or bio weapons work that makes it even more critical that our Office of Countering Weapons of Mass Destruction is reauthorized by Congress urgently, and ensuring that AI cannot be used to generate child sexual abuse material and other information that threatens Homeland Security,” he said. “We’ll be partnering with industry and experts to advance those discussions. And then finally, it looks at promoting cybersecurity in the AI space, overall work that extends what CISA has been doing incredibly well since it stood up.”
DHS’ long-standing AI efforts
DHS listed 45 use cases in its AI inventory posted as of Oct. 30. Hysen said some of the current AI uses cases include the Customs and Border Protection using predictive models to better target inspection resources to identify vehicles that may be hiding drugs or other dangerous materials coming into the country, or FEMA using AI to support a faster ability to identify damages from a disaster to get aid to citizens.
Along with naming Hysen as the department’s first chief AI officer in September, DHS also has issued new policies to more quickly and “surgically” prioritize areas where the department can adopt the technologies, specifically around facial recognition and face matching, and created an AI Task Force, which is charged with defining the goals and strategy for the use of AI.
“I expect that over time, the percentage of IT that we manage at DHS, and really any large enterprise that manage and uses AI, will only grow and will approach 100% in various forms, and so effectively and responsibly leveraging AI is going to become a critical part of any IT organization moving forward,” Hysen said in an interview with Federal News Network. “We are focused on being able to build out procedures that ensure that our use of AI across DHS is safe, ethical, responsible and effective, that we are rigorously testing to ensure that our uses of AI do not demonstrate unintended bias, and that our AI use is explainable to the people that we serve. That’s something that we’re doing in close partnership with our Office for Civil Rights and Civil Liberties, our privacy office, our general counsel and many others across the department. We are doing that in a methodical way.”
Hysen said for generative AI, DHS is going through a similar process to ensure that it is using the technology responsibly.
DHS to launch IT Academy
A key piece to that responsible use is workforce training. Hysen said DHS is taking both a specific and generalized approach to AI training.
“The first is how we train our IT professionals. Across the department there are over 5,000 people we want to be able to build skills around AI data science and related fields to manage software acquisitions and other complex work in this space. That’ll be done as part of the new IT academy that we are standing up across DHS,” he said. “The second is around AI literacy for the entire DHS workforce. We are still in the early stages, but we are looking to offer AI literacy training to every DHS employee so that they can understand how to use AI systems effectively. They can understand risks posed by unintended bias or hallucinations, and know what how they should be thinking about using the outputs from AI in their work.”
The CIO’s office has been developing the IT academy and related courses over the past 18 months. Hysen said he expects the initial pieces to be available in early calendar year 2024, which is a standard new hire orientation program for all new IT employees anywhere at DHS. He credits acting DHS CFO Stacy Marcott for this idea, which her office uses for new financial and budget employees.
“It helps build from early on a sense of cohesion among the workforce. The second is that we are looking at a standardizing an entry level program. That’s a part of our DHS cybersecurity service or cyber talent management system, to be able to have a standard curriculum for entry level employees to learn about different IT fields and be able to start their careers off impactfully at the department,” Hysen said. “Then finally we’re looking at ongoing educational programs and offerings for all employees to build new skills in fields like AI and data science, customer experience and design and the like.”