IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

To Buy AI, Government Needs to See Big Changes — Fast

At a recent event bringing together public- and private-sector leaders in government technology, many spoke about how hiring struggles are making AI both more appealing and more difficult to adopt.

From left to right: Peter Pirnejad, city manager of Los Altos Hills, Calif.; Vivian Nguyen, city councilor for Everett, Mass., and co-founder of Legislaide; Parth Shah, co-founder and CEO of Polimorphic.
Photo by Ben Miller.
Want to sell AI to government? Make it as easy and helpful as possible.

That was the takeaway from government technology leaders in both the public and private sectors who gathered recently at the State of GovTech 2023 event in Foster City, Calif. Across the one-day event, speakers shared the pressing needs to make up for a national struggle in the public sector to hire and retain qualified IT workers.

Ironically, AI holds a natural promise to help governments to address labor gaps, but because of those gaps, decision-makers lack time and resources to evaluate the technology.

“We use the analogy of an iceberg; 80 percent of what we do is under the waterline, meaning most of our resources are just keeping the lights on,” said Peter Pirnejad, city manager for Los Altos Hills, Calif. “Getting the permits issued and making sure that the water pipes are fixed, making sure that capital projects are properly managed and implemented — all that takes 80 percent plus of our workforce. So that only leaves about 20 percent of your workforce to do those new innovative digital transformations or modernizations or experimentations or pilot programs.”

So for local leaders such as Pirnejad, the bar is high: They need solutions that make a big difference without taking a lot of time or resources to implement.

“It's really important that if you come to a local agency and you're trying to pitch your product, it's got to be A to Z, it's got to be the entire package,” he said. “It can't just be a very cool, innovative solution. It's got to be, ‘We'll help you integrate it, we’ll help you train, we'll help you pass it on.’ And so if you have the solution that is able to get you up to speed quickly, it's scary helpful. But also you have to realize that we don't even have the time to watch the demo, let alone work on the transformation.”

Those “extras” — staff training, implementation, etc. — are important because they help prevent the solution from becoming a burden instead of a boon once the agency has begun using it.

“You can't be like, ‘Hey, it takes 50 steps to do this, and it takes only 45 steps to do ours,’” said Anthony Jamison, co-founder and CEO of gov tech accelerator and event co-host CivStart. “So it needs to be something drastic for them to make that change, they have to really come in and educate and skill up and staff their team to be able to utilize this whole new service that they're replacing their own service with.”

It helps, Pirnejad said, when AI is working with a relatively narrow data set. That’s because of “hallucinations,” the term for when AI produces outputs that IBM calls “nonsensical or inaccurate.”

“The big challenge is around wrong data — the hallucinations that it might prepare, so we are specifically looking … for uses that have a very defined data set. So if the data set is the entire Internet, you're gonna hallucinate a lot. But if the data set is very defined, it's a video of a council meeting and you're trying to run minutes, or if it's a website that's controlled and you're trying to do searches on that, I think you're less likely to get errors.”

Companies trying to sell AI to government might also do well to remember that not everybody is comfortable with the technology just yet. Vivian Nguyen, a city councilmember in Everett, Mass., and co-founder of the gov tech startup Legislaide, said many in her municipality are hesitant to even try AI solutions. However, they’re much more open to hearing about solutions to the problems they’re having — regardless of the technology supporting those solutions.

“I feel like we're at a point where these responsibilities are building up and we're not able to catch up and they aren’t resistant to technology, they're just resistant to how you deliver your solution,” Nguyen said. “And I think that's how you should go about it, make sure that you are a practical solution for them and delivering it in a way that appeals to them, because they don't want to hear about AI. They don't like how fast the world is changing. You just need to make sure you message it as, ‘Hey, I have the solution. This is going to make your process a lot more effortless. It's not going to take a lot of knowledge, you don't need to know how to use technology.’”
Ben Miller is the associate editor of data and business for Government Technology. His reporting experience includes breaking news, business, community features and technical subjects. He holds a Bachelor’s degree in journalism from the Reynolds School of Journalism at the University of Nevada, Reno, and lives in Sacramento, Calif.