Here’s where tech policy leader Ed Felten sees a role for AI in the federal government

With President Joe Biden‘s massive infrastructure bill gaining traction, there’s plenty of talk in the federal government about the role of tech in everyday operations. Ed Felten, director of Princeton’s Center for Information Technology Policy, hopes that can mean a revamp for federal AI policy in the coming months.

“Tech can help the government do its jobs, do its mission more effectively,” Felten, who is also the host of the recently-launched podcast A.I. Nation, told Technical.ly. “Government can have better visibility into what’s happening, more ability to shape and target the programs that are going on.”

Felten, a current member of the executive branch’s Privacy and Civil Liberties Oversight Board and former Deputy U.S. Chief Technology Officer during the Obama administration, said it’s still too early to know how the Biden administration will bring AI into its overall strategy. But things are already different than they were under former presidents Barack Obama and Donald Trump. In his time in the White House in 2015 and 2016, he said everyone was just getting started on understanding AI and developing a strategy. In the years since, he’s seen AI grow in industry and an uptick in uses, such as in facial recognition software.

“My sense is that the use and development of these technologies has advanced faster than the conversation about them in those years, certainly within the government,” Felten said. “I would expect the Biden administration to be paying a lot of attention to this issue as they start to build up their tech policy strategy.”

Even if it’s still getting started, Felten thinks that there are plenty of applications for AI in government operations. It has already made its way into our everyday lives (think social media algorithms, self-driving cars and replacing jobs with robots). At the federal level, though, it could be a solution for the challenges that come with the federal government’s size. AI can find out who knows what, or where to find new information, potentially shaping policy through a more comprehensive approach.

Ed Felten. (Courtesy photo)

“Just knowing what’s going on and who within the federal workforce can answer a question or has domain knowledge that’s useful, that can be really challenging,” Felten said. “So to the extent that AI helps the government communicate with itself better, helps to find information that it has, helps to find [new] information, helps to get more actionable knowledge out of the data sets that is has, that can also be very useful.”

To Felten, the best way for the federal government to get AI immersed in this way is via in-house technologists. He said one of the biggest mistakes agencies can make is thinking of technology as just something that can be bought to fit an organization’s needs. Without hired experts, he said government might have trouble understanding the best software to buy, or even how to operate something after purchase.

“You really do need tech talent that is integrated with the mission parts of an agency or organization that are working on the mission,” Felten said. “To not have enough tech talent inside an IT department is really a recipe for missing the technical opportunities that exist.”