The hearing featured various speakers, including Leo Schoonover, the state’s chief technology officer for Florida Digital Service; Sachin Koshy, a senior AI strategist for World Wide Technology; Traci Lee, head of state and local government affairs for OpenAI; and Edward Longe, director of national strategy at the James Madison Institute.
The main topic of discussion among panelists and lawmakers was the state’s use of AI and what it might mean for government services in the future.
Currently, the technology is being used within several state agencies. Schnoover said the Agency for Healthcare Administration has developed an internal AI model for Medicaid analysis, while the Department of Environmental Protection is considering AI for assistance with the Florida Geospatial Open Data Portal.
The Division of Emergency Management is also using AI to detect invoice discrepancies, and the Florida Department of Revenue is implementing robotic process automation to help residents fill out certain tax forms and power chatbots to help navigate child support processes, he added.
In general, though, each agency independently oversees the technology, most of which is being used internally.
“They don’t come to [FLDS] for permission to move out on their internal operations and development,” Schoonover said. “We are aware, and a lot of the ones that I’m aware of are internal.”
Furthermore, he explained that Florida Digital Service, which leads the state’s enterprisewide cybersecurity, data interoperability and cloud-first initiatives, has standards that it monitors, and keeps a pulse on the agencies that need help.
However, “for the most part, they’re independent, developing their own AI strategies,” Schoonover said.
As for technology concerns, a couple of lawmakers asked questions about AI’s impact on the state’s open records policy and workforce concerns.
For example, regarding open records, Rep. Fiona McFarland, R-73, shared that she’s spoken with several local governments that have expressed concerns regarding queries input into AI becoming subject to Florida’s Sunshine Law, which makes any records made or received by state or local government agencies public record.
Lee, however, explained that OpenAI’s government products, such as ChatGPT Enterprise and ChatGPT Gov, have special administrative controls such as single sign-on and other ways to comply as additional protections.
“Our ChatGPT Gov product, which was recently announced, allows government to self-host within its Azure cloud environment,” Lee said. “The types of deployments we’ve seen are both short-term pilots, where government, let’s say, wants to see how effective our tools may be for operational efficiency ... [and] larger scale deployments within government, but they’re mostly using our Enterprise and our ChatGPT Gov tools, which have additional protections on top of the consumer-facing product.”
From a government perspective, Schoonover suggested implementing record exemptions similar to those used to protect the state’s cybersecurity efforts from public records.
“If we start to leverage artificial intelligence to do analysis against cybersecurity incidents, I think we should be able to still exempt those from public record, especially if it gets the details of vulnerabilities where we’re being exploited potentially or, you know, maybe our tactics, techniques and procedures that we develop internally while using AI,” he said.
As for workforce concerns, Rep. LaVon Bracy Davis, D-40, expressed concern with AI replacing state employees.
“I’d love to explore more how AI can work in tandem with the state,” Davis said. “I’m not in agreement with replacing bodies and mass firings and all of that, but I am in agreement with AI helping to be more effective and efficient.”
Longe explained that initial estimates of 45 percent of job elimination due to AI have now dropped to about 5 percent within the U.S.
On the other hand, Koshy explained that workers might be replaced by those who know how to use AI effectively instead of having technology replace workers directly.
Concerns aside, panelists said the technology could be used in the following ways within state government:
- Longe said AI could be used for re-employment assistance process improvements; streamlining regulatory frameworks; detecting waste, fraud and abuse; and reducing administrative tasks for state employees.
- Lee said AI could be used for HR process classification and evaluation, enterprise translation services, constituent letter categorization and transcribing handwritten constituent letters.
- Koshy expressed that AI could be used for public safety improvements, video surveillance analysis, contact center and help desk efficiency and procurement process optimization.
A full recording of the meeting can be found online.
