Use of Artificial Intelligence Poised to Grow in State Government, Survey Finds

Artificial intelligence is seen as the emerging technology with the most potential by state chief information officers, yet basically none say AI is widely deployed in their state, according to a new report.

The findings detailed in a report from the National Association of State Chief Information Officers, Center for Digital Government and IBM underscore the lengths states have to go to realize the potential of artificial intelligence in state government.

Only 1% of state CIOs said the technology is “widely used” in their state while another 19% said they are piloting artificial intelligence, 13% are using but not in “core lines of business,” 31% are engaged in demonstrations or proofs of concept, and 24% are evaluating proposals.

“A lot of activity is in the very early stages of development,” said Joe Morris, vice president of research at e.Republic, as he spoke about the report’s findings at NASCIO’s annual conference in Nashville this week.

Forty-five states provided responses to survey questions analyzed for the report. The biggest hurdle identified by state CIOs in implementing artificial intelligence is the ability to overcome the constraints of their current information technology infrastructure.

“Most legacy environments weren’t designed to handle the large volumes of data and processing that advanced analytics or AI applications demand,” the report said. “That leaves organizations scrambling to upgrade their on-premises data centers, or more likely, to develop hybrid cloud strategies that can provide the necessary capabilities.”

Sixty-five percent of state CIOs view artificial intelligence and robotic process automation as the most impactful emerging technology in the next three to five years, according to NASCIO’s 2019 survey. 

While states have ideas about the areas in which they’d most like to put artificial intelligence to work, few have fully developed policies or guidelines related to oversight of the technology.

Cybersecurity, fraud detection, digital services for citizens and traffic management were the areas where the survey found states most highly anticipated to use artificial intelligence. Yet only 9% of those surveyed said they had a policy regarding the responsible use of artificial intelligence.  

The report says most of the current use of AI by states is with chatbots and digital assistants on websites, which can be used to support IT help desks in answering calls about common technology problems, with one out of four respondents saying these tools are in use in their states.  

Utah’s chief information officer, Mike Hussey, is among those who have sought to broaden the way the state utilizes artificial intelligence, embracing opportunities to experiment.

The state had sought to use the state’s Department of Transportation cameras as a way to spot and identify wildfires and vehicle fires, but quickly learned the machine learning algorithms used to identify these images wouldn’t work on their older camera systems. Rather than give up, Hussey said the state began an initiative that would allow the algorithms to work with the older cameras.

Hussey said he is also thinking of expanding the use of digital assistants or website chatbots to increase residents’ access to information online and relieve staff of routine and time-consuming tasks.

“Some of the value is just dipping our toe in the water and more of an experimental thing with some of these digital assistants to see if we can engage the citizenry on this platform,” Hussey said. “If we are finding value there how can we transform it into more of a transactional type of system? For instance, you might say ‘Alexa, renew my business license’ and it’s very Amazon-esque and it walks you through the process very simply.”

Hussey said government has a responsibility to reach the entire spectrum of residents, and can improve engagement with millennials and other technology-savvy resident through these types of initiatives.

“We are trying to keep people out of line, rather than in buildings,” he said. “There is a target rich environment in terms of trying to simplify some of the processes we have.”