AI faces growing pains in the federal workplace

Auditors want artificial intelligence tools to monitor fraud and abuse, but the new tech is slow to penetrate agencies.


More and more, federal agencies are looking to integrate artificial intelligence and automation into their day-to-day workplace functions, particularly to implement internal controls and compliance measures to counteract fraud and improve efficiency.

At a Sept. 18 Association of Government Accountants (AGA) event, experts in AI and robotic process automation from public- and private-sector organizations stressed that automating lower-level functions, such as data processing claims, would free up employees to work on higher functions that require more analytical skills and mission-specific duties.

"Artificial intelligence operates in real time to enable personalized experiences," said Dan Chenok, executive director of IBM’s Center for the Business of Government and a former government official. Comparing AI to the chat boxes that often accompany internet users when they’re shopping online, Chenok pointed out that such technologies are constantly building upon themselves to improve users' experiences and stamp out common problems that users may encounter when using a government website.

AI and RPA also allow workers to identify patterns of potential fraud or other misuse, according to Jennifer Main, the chief operating officer for the Center for Medicare and Medicaid Services.

CMS processes over 900 billion claims per year. Although it employs over 75,000 contractors in addition to its 6,300 permanent employees, fraudulent payment claims such as "double dipping" are common, said Main. Using aspects of AI in the algorithms CMS uses to track claims have allowed it to identify trends as health care costs rise along with "where they’re rising and why we're paying more in certain [areas]."

However, issues with AI and RPA persist, particularly among career employees who are resistant to new forms of technology in the workplace or fear that new technologies could mean loss of jobs.

Erica Thomas, the RPA program manager for the DOD’s Comptroller Office, said that engaging employees early on in the process and developing buy-in can mitigate that.

Even so, if a real-time poll of AGA training attendees is anything to go by, AI has a way to go to make an big impact in the daily lives of federal employees: 71% said that there was no AI or machine-learning project in their office, and 52% reported that there were efforts to begin developing on in the next three years.

"When computers were introduced into the workforce, however, many years ago, there was a lot of concern that these are going to replace people's jobs," Thomas said. "Now when you look around, people have different jobs, they have different things to focus on, they’re more efficient. So, I don't view RPA as a replacement factor at all, it's more of an upscaling and redirecting resources to appropriate tasks."