For most states, AI adoption is still in early stages, Code for America report shows
A report published Tuesday by the civic tech nonprofit Code for America reveals that while many states are interested in using artificial intelligence to improve digital services, most are still in the early stages.
The Government AI Landscape Assessment report found only three states — New Jersey, Pennsylvania and Utah — have made major strides in building their AI capabilities, by hiring tech experts, training workers and upgrading outdated systems.
Of the remaining 47 states, the report determined, most are in the “developing” stage, still experimenting with AI projects, and often running small pilot programs or relying on outside partnerships with universities or technology companies for help.
“Across 50 states, I think people are realizing that AI is something that needs to be considered in service delivery,” Jenn Thom, lead data scientist at Code for America, told StateScoop in an interview. “And while there are some states who have made progress more quickly, I think those states can serve as a model for states who are just dipping their toes in or starting to explore.”
Last year, New Jersey appointed the state’s first chief AI strategist. Utah established the Office of AI Policy, one of the only state agencies in the nation solely dedicated to AI governance. The Utah office also partnered with the Aspen Institute’s Policy Academy on a yearlong report designed to help state governments craft responsible AI policy.
Thom said that in addition to hiring personnel who are focused on AI, these states also have strong data infrastructures and shared IT systems, which also helped to accelerate their AI adoption.
“Pennsylvania has a centralized IT organization, so I think that has really helped them move a little more quickly in adopting some of and adopting and getting some of these AI pieces of infrastructure up and running,” Thom explained. “Utah also has already a really good cloud environment, and so I think that was something that actually helped them be able to adopt AI more quickly at scale as well.”
The report stresses that while AI has the potential to make government work faster and more fairly — for example, by helping people apply for benefits more easily or processing paperwork more efficiently — it also warned that states need strong guardrails to ensure AI tools are used responsibly and don’t harm vulnerable communities.
Pennsylvania Gov. Josh Shapiro in January announced a new pilot program in collaboration with OpenAI that will provide licenses for some state employees to begin using the enterprise version of ChatGPT. Employees are using the technology to, among other things, ensure agencies are properly aligning their hiring needs with the state’s job classifications.
“That’s an example of something that allows the people on the ground doing the work to be able to adapt the technology to the problems that they’re encountering every day,” Thom said of the pilot program.
The report recommends that states exploring AI start small, learn from early experiments and focus on using AI in ways that serve people — not just adopting technology for technology’s sake.
“AI as a technology has a lot of potential to transform how services are being delivered within state governments, but it needs to be deployed in a human-centered way,” Thom said.
