States wrestle with AI-driven future

Two states, on two sides of the country, illustrate some of the uncertainty that state IT leaders feel over the role of artificial intelligence, machine learning and generative AI in government operations.

California Gov. Gavin Newsom called for the state to preserve its role as a “global hub” for generative AI and a leader in “shaping the future of ethical, transparent and trustworthy AI.” A recent executive order on the technology, requires agencies to draft a report by early November on potential use cases, risks and policies to guide the state government’s use of generative AI. 

Meanwhile, Maine in June instituted a six-month moratorium on its state government’s use of the technology, saying the pause—that could be extended—is necessary to help keep the state ahead of a “rapidly evolving cyber threat landscape.” 

Numerous other states and cities have issued guidelines on generative AI’s use, all with an eye on the evolving technology.

Amid that uncertainty, state IT leaders gathered in Minneapolis for the National Association of State Chief Information Officers annual conference. They were intrigued by some of the potential use cases for generative AI and machine learning, but are still unsure how to proceed. Indeed, California CIO Liana Bailey-Crimmins said that agencies are already experimenting with products and services that use AI, there is still a lot to learn.

“Believe it or not, the IT community still hasn’t figured out the difference between traditional and generative AI,” she said during a panel discussion at the conference.

There is plenty of excitement surrounding the technology, too. Of those who responded to NASCIO’s 2023 State CIO Survey, 53% said generative AI will be the most impactful emerging IT area in the next three to five years, with another 20% citing AI and machine learning more broadly.

Still, some state leaders urged continued caution on embracing generative AI. Vermont Director of Artificial Intelligence Josiah Raiche said during a conference workshop that governments should implement codes of ethics before devising policies governing their use of AI. Raiche, who is one of just two state leaders with that job title, said he aims to build a coalition of state AI directors—as more are hired—in a bid to encourage “good governance” of AI across the states.

Other state tech leaders are pushing for a consortium-based approach to help agencies get up to speed. Washington CIO Bill Kehoe said during a panel discussion that states that are ahead on generative AI should share their findings, use cases and best practices. That way, he said, it can be more than just an “academic discussion.”

Education on generative AI will be crucial, North Dakota’s Deputy CIO Greg Hoffman said during a panel discussion, especially as it promises to be “that next huge turning point from a technology perspective.” It will change how agencies work and how citizens and users consume services, he said.

And state tech leaders repeated their confidence that AI is not coming to replace employees, but instead will augment certain functions of their jobs and help agencies deal with a shortage of workers, all while ensuring humans make the final decisions.

“Most states are in that predicament where you have a set number of staff that you're allowed to bring on through appropriations,” Hoffman said. “Beyond that, as your needs increase, as demand increases, how do you meet that demand? There isn't always the opportunity to just wait and set something aside for another cycle to get that workforce.”

That’s where AI can help. 

Some states have already gained efficiencies using AI. North Carolina Chief IT Procurement Officer James Tanzosch said during a breakout session that the technology has helped streamline procurement by eliminating repetitive form filing by government and vendors and helping procurement officials ensure they have the right documents. In time, he said generative AI could help write requests for proposals, albeit not the whole document, but perhaps get RFPs “80% there.”

AI-augmented cybersecurity is a use case that especially excites state IT leaders, as they believe it can enable quicker response times to attacks and help find vulnerabilities. Montana CIO Kevin Gilbertson said states will need generative AI to combat hackers, who also will embrace the same tools.

“The traditional way of detecting threats based on signatures and then human responses, that's all going away very, very quickly in favor of automation,” said Ian Milligan-Pate, area vice president of state, local and education at security company Zscaler. AI can also help governments that traditionally do not have a large cybersecurity staff do their own monitoring, he said during an interview at the conference.

Meanwhile, state governments continue to wrestle with fraud, such as the theft of millions of dollars from pandemic-era unemployment insurance programs. These threats will continue as hackers get more sophisticated and embrace emerging technologies, like generative AI.

Generative AI is also fueling disinformation. Fraudsters can “create images, voices, videos that can pass [the test of] the human eye,” Matthew Thompson, senior vice president and general manager of public sector at identity management company Socure, said in an interview. When people trust a government “document, a voice, a video, less and less,” it weakens trust in government, he said.

States must learn how to leverage AI to defend themselves against increasingly complex and sophisticated threats, he said, and ensure that laws and policies enable them to truly embrace the technology and not be left behind.