Nevada is using AI to prescreen unemployment claims

For the last few months, Nevada’s unemployment agency has been using a new artificial intelligence to prescreen claims, a change that has accelerated the approval process 30-fold, according to Timothy Galluzi, the state’s chief information officer. 

In a recent interview with StateScoop, Galluzi said the state’s Department of Employment, Training and Rehabilitation is cutting through its unemployment claims backlog by allowing an AI tool to handle some of the process’ most tedious aspects. He said the tool has managed a 99.99% degree of accuracy, but two senior analysts still have the final say on approving or denying each claim.

“What this has done is sped up the review process and it’s in keeping with the tenets of our state AI policy,” Galluzi said. “It has kept a human in the loop in our review process, it is secure, it is kept compartmentalized, personal identifiable information is secure, personal information is not getting released to the public.”

Galluzi’s office, which is a Cabinet agency under Gov. Joe Lombardo, published the state’s AI policy last November. Similar to policies in other states, Nevada’s document defines terms and rules, sets a governance structure and data protocols, and contains principles to guide the state’s use of generative AI and other automation software.

Nevada’s six AI principles include fairness and equity; innovation; privacy; safety and security; validity and reliability; and transparency, accountability and explain-ability. The five-page policy document also includes explicit prohibitions on using AI to do things like “engage in deceptive practices” or “infringe on privacy rights or data protection laws.”

Galluzi said the tool, which staff have limited exposure to, began with “a little bit of trial and error,” but he believes its success reveals how ready state governments are to use AI for a host of functions.

“I think states are really primed for leveraging AI technologies because we have very repetitive processes,” he said. “We’re built for repetitive processes, because we have governance, we have policies, we have a lot of repetitive tasks. And AI is uniquely positioned to be able to automate repetitive tasks.”