Government leaders see ‘momentum’ in agencies’ AI adoption

LAS VEGAS — Twelve months ago, state and local government agencies were cautiously piloting artificial intelligence, exploring limited, low-risk uses for how it could boost employee productivity.

One year on, speakers and experts at last week’s Google Cloud Next conference in Las Vegas said interest in AI has continued to spike across governments and moved beyond those pilot programs into full-scale implementation.

“The biggest excitement we're having is the conversation has moved from pilot, pilot, pilot,” Matthew Schneider, Google Public Sector’s go-to-market leader for U.S. state, local and education, told Route Fifty in an interview at the conference. “Last year, everything was about, let's try something. Let's figure it out. We are seeing AI now being implemented, whether it's at the state level, at the city level, education especially. I feel like the momentum behind it is as fast as, if not faster than, what we're even seeing from some enterprise customers.”

A February report from Google Cloud and the National Research Group showed how much momentum is behind government AI, including its agentic use. According to the report, 55% of leaders in the public sector are using agentic AI in their organizations, with 42% saying they are using more than 10 AI agents. The majority reported improved employee productivity because of the tools, as well as faster response times to cyber threats.

“We're now starting to feel like there's a bit of momentum around an understanding of ‘where do we trust these current AI agents to have access to tools, have access to resources,’ and take it to as far as a logical conclusion as you feel good about before you need to move to a human in that loop,” Chris Hein, Google Public Sector’s field chief technology officer, told Route Fifty in an interview at Google Cloud Next. “It becomes much more than just a question-and-answer chatbot.”

For public sector leaders, getting to the point where they are using AI beyond a limited pilot project has taken a lot of hard work. One leader on that has been New York, which has developed AI through several dozen use cases in an effort that Chief Information Officer Dru Rai compared last year to “science experimenting.”

Since then, he said last week, pilots appear to have been successful, as 87% of the employees surveyed across the six agencies that used AI said they want to use the technology more in their work. It will now roll out to several hundred thousand state employees, and all the while, New York has been refining its AI policies since first publishing them 18 months ago. Doing so reflects how the technology and expectations of it have evolved since then, Rai said, and helps get more employees comfortable using it in their day-to-day work.

“The definition of AI is changing,” Rai told Route Fifty in an interview at Google Cloud Next. “The line between [software as a service] on-prem software and AI is kind of blurring, and now it's more about what's the output? How do you get your output? If you get your output using inference and some algorithm which is pre-trained, or anything which is artificial, then we have to define that, so we are working towards changing the policy.”

Rai said New York may appear “slow” compared to some other states in adopting AI, but that was partly due to the need to ensure the technology was secure, so the state took several months to make the necessary adjustments. While any uncertainty around how secure an AI system is may worry some government leaders and prevent them from moving beyond a pilot project, Rai said agencies can manage and mitigate risk while at the same time embracing innovation.

“The governor [Kathy Hochul] and I believe that the best way for AI is not to run away, not to fight it, but to understand, use, find the problems and help fix it,” Rai said. “Take the good part of it, apply it more and just do more with less. We live on taxpayer money, so if we can do more with whatever we have, that's great.”

Others agreed that jumping in and being bold helps move an AI implementation away from just being a pilot project. Los Angeles Chief Information Officer Ted Ross said pilots can sometimes feel like “standing on the sidelines watching the game,” and while governments must still be cautious, they have to embrace where the technology is going.

“We are seeing a massive shift in how computing is done and how user interfaces are done,” Ross told Route Fifty in an interview at Google Cloud Next. “We, as the city of Los Angeles, can't stand around watching the shift. We have to participate in it. It doesn't mean that we should move fast and break things; that would be highly irresponsible for government. However, it does mean that we need to engage a lot of our constituencies. We need to get them trying out tools and products. We need to identify where the value is, and we need to start enabling that value and the strength of this.”

Leaders are already encouraged by what they are seeing, especially in ways that look to make people’s lives better. And they are seeing dramatic improvements in the technology itself, which helps encourage its further adoption. Hein noted that a year ago, a lot of the talk was about AI hallucinations, which could “stop a pilot in its tracks.” Now, he said, the models and data are improving.

Austin, Texas, for example, has turned to AI to help bolster its tree canopy and protect residents and visitors from extreme heat, and those uses are among the most exciting for those involved.

“At the end of the day, we strip away all the technology, strip away all of this fancy jargon we use, this is affecting people's lives,” Google Public Sector CEO Karen Dahut told Route Fifty at Google Cloud Next.