Best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews on Apple Podcasts or PodcastOne.
In the months following President Donald Trump’s executive order prioritizing the development of artificial intelligence in government, the administration has also taken steps to ensure the federal workforce sees AI as an asset rather than a threat.
But what if artificial intelligence could predict when agency employees feel like quitting before they’re out the door?
Ranjeev Mittu, the head of the information management and decision architectures branch of the Naval Research Laboratory’s IT division, speaking Wednesday at an FCW conference on AI, said the lab is looking at using AI tools to comb through data from exit surveys and to flag common workplace issues.
“As employees are progressing in their career, we can look at the predictors. Do those predictors tell us we might have an employee that is going to leave? Should we fix a problem because a lot of employees are leaving for a given problem that no one’s addressed? I think there’s a rich opportunity not just to apply machine learning, but a variety of approaches under AI to solve this kind of problem,” he said.
However, Mittu said while AI will gradually reshape the federal workforce, that rate of change will likely happen at a “slower pace” than what’s happening in the private sector. But several agencies, including the General Services Administration and the IRS, have already found robotic process automation to be a valuable time-saver for its employees.
By leveraging RPA, Mittu said one naval command reduced the time spent on manual tasks — like copying data from one spreadsheet to another — down from three weeks a month to about four days.
“It allowed this analyst to focus on other problems within the command, other improvements in the business activity that were not being addressed, because now the manual stuff that really was really brainless, we automated that process,” he said.
The Trump administration’s focus on artificial intelligence has also driven the need for more high-quality data, which Tim Persons, GAO’s chief scientist and managing director of its Science, Technology Assessment, and Analytics Team, said requires a different data management approach from agency leadership.
“Moving from a CIO mentality, which often has to build the roads and bridges to keep data moving around, we need to move toward this chief data officer [mindset], what you can do with that data; an asset, not a burden to move around,” Persons said.
While the quality of government data has proven to be a stumbling block for AI pilots, GAO’s Science, Technology Assessment, and Analytics Team has set up a data “sandbox” within the agency to work with unstructured data sets.
By working with the unstructured data in the sandbox, Persons said GAO looks to improve data collection standards and reduce the time spent sifting through “noise,” or invaluable data.
“When you have a data-centric model, you’re going to deal with mess, and you can actually have a high rate of failure,” Person said. “The hope is that you when you get in the mess, you learn from the mess, and then you start to adapt it and tweak it.”
The current state of AI technology, Persons added, has also reduced the level of expertise needed to work with AI tools.
“You don’t have to be a Ph.D. in math and statistics to survive in this world,” he said. “You just need to have a basic element of things.”
Federal employees’ attitudes around AI have also shown improvement. In a February study of more than 500 federal employees, Accenture Federal Services found that 51 percent of respondents said they expect an increase in the share of roles requiring collaboration with emerging technology, including artificial intelligence, over the next three years.
“There is this perception that this is actually a benefit to them, that it will create opportunities for them to continue growing and developing,” said Michael Gavin, a senior manager for human capital at Accenture Federal Services.
However, the study also found that 73 percent of respondents said their agency’s leaders have not done a good job communicating a long-term vision for what AI will mean for their workplace.
“So clearly there’s a disconnect here, around some of the enthusiasm that workers are feeling and then the expectation that they should be hearing more,” Gavin said.