Fairfax County to educate and train its employees on using generative AI

As the use of chatbots and other types of “artificial intelligence” continues to grow, Fairfax County is working to ensure its government employees are educated on the opportunities and risks associated with the technology.

During an information technology committee meeting on Tuesday (June 4), the Fairfax County Board of Supervisors received an update on the use of generative AI within the county government — including how to “deliver improved business outcomes” while maintaining the county’s “rigorous security standards.”

Maura Powers, the county’s Department of Information Technology (DIT) program director, said the department worked with the Information Technology Policy Advisory Committee (ITPAC) to develop employee guidelines for using AI, which were released this March.

Power said the guidelines ensure county employees understand the risks of publicly available generative AI tools, such as ChatGPT, the popular but controversial virtual assistant by OpenAI that produces text, images and audio based on pre-existing — and sometimes copyrighted — data.

“We wanted to make sure we maintain that security posture that we’ve always had, making sure that we’re building processes in place, because now artificial intelligence is coming baked into most of the products that are out there,” Power said.

The DIT also conducted research to determine best practices and responsible uses for generative AI.

“We’ve looked at different AI technologies and what we can use to provide a secure method for generative AI use for our employees and for the public,” Powers said. “We’ve looked to make sure that any of the systems and vendors that we work with have responsible AI use policies.”

Powers highlighted instances in which AI tools created by and contained within the county have helped improve the efficiency of employees.

“One that I think is very cool and interesting is where we’ve used it for data analytics reasons. So, we’re using the power of that large language model against the databases that are contained within our enterprise,” she said.

The county also has an exclusive general chat assistant that utilizes AI technology.

“There are products on the market that allow you to do the same thing. What I think sets our chat assistant a little bit apart is that it’s closed,” Powers explained. “It’s contained within the Fairfax County environment, so no one else owns that data.”

On the Fairfax County Public Schools side, ITPAC member Dr. Rajni Goel, who’s also a Howard University professor focused on cybersecurity and information systems, emphasized the need to ensure that students and faculty within the county understand AI.

FCPS officials and the school board have discussed developing guidelines for the use of AI, including in classrooms, but no policies have been officially adopted.

“K-12 education and the leadership there, really, I believe it should work hand-in-hand, interlocked with what our county is doing,” Goel said.

Some supervisors noted the importance of training employees on responsible AI use and security. County Executive Bryan Hill said they are working to consider how to implement countywide training.

Hunter Mill District Supervisor Walter Alcorn said the guidelines on using AI were a great first step and suggested that, next, they should be adapted to individual agencies, rather than a centralized set of standards.

DIT Deputy Director Nate Wentland assured the committee that they are looking into an individualized approach.

“We’re really trying to have the agencies in the business help to drive those next steps,” he said.

Springfield District Supervisor Pat Herrity, who chairs the IT committee, asked ITPAC to come back to the board with a plan for implementing employee training and education.

“I think there are a lot of employees out there afraid of it,” Herrity said. “They’re not going to jump into it unless we give them the comfort level…within their own work world, not system wide.”