Across the country, state legislators are beginning conversations about how AI, and specifically generative AI, can help state agencies better serve their residents. Amid a national environment of shrinking state budgets and growing demand for technologically advanced workers, AI promises new solutions for state governments to help address these issues. But with these AI opportunities come challenges. How are state governments meeting these challenges? The State of Wisconsin offers good guidance.
The Promise and Potential of AI
Many states are facing the challenges of a declining workforce and the need to conform to strict budget rules that limit the state’s spending. As unlike the federal government, most states are required to produce a balanced budget on an annual basis.
To address these issues, Wisconsin and states across the country are beginning to explore ways to introduce AI into their government operations to drive new efficiencies. For example, some states are looking to generative AI for policy analysis, helping legislators and government employees better understand the impact of lengthy, complicated legislative bills. Some states are planning to use AI in workforce planning to analyze and more effectively optimize government workforce and other resource allocations. In law enforcement and public safety, some states are implementing solutions to streamline criminal case management. And states are also seeing the value for citizen services, such as deploying AI in the form of chatbots and virtual assistants to provide 24/7 support.
Define AI
At least 34 U.S. state governments have launched studies on AI policy, but only a dozen state governors have issued executive orders on AI policy, according to the nonpartisan policy-research group MultiState. Wisconsin is at the forefront of efforts to define the potential use and benefits of AI by state agencies. In 2023, Governor Tony Evers signed Executive Order # 211 which created the Governor’s Task Force on Workforce and Artificial Intelligence. In July 2024, the state released the task force’s Advisory Action Plan. It offers a series of policy proposals for implementing AI across four key state government functions: education, government, the workforce, and economic development. Concurrent with the Governor’s Report, a special Speaker’s Task Force on Artificial Intelligence in the state legislature is tasked with defining AI, and exploring how it is currently used and should be used in the state. The parallel initiatives showcase the importance of defining the terms of AI engagement in advance of moving forward with any major AI legislative initiatives.
Challenges: Lack of Understanding
One of the most crucial tasks is educating stakeholders on the value and responsibilities of AI policy and legislation. The Governor’s Plan identified the need for additional work to “develop guidelines around best practices in AI implementation and use, including ethical and equitable decision making, risk assessments, and policy/laws governing social impacts.” The Governor’s Task Force also found that to move the actionable ideas forward, “more specific policy and program expertise would be needed,” and that the state needs to “use strategic communications and outreach to foster public trust.”
Demonstrate AI’s Value
In parallel with the Governor’s approach, the legislative task force members emphasize the need to work proactively to educate their colleagues and change negative perceptions of AI. One strategy members have recommended is to showcase successful examples of industry adoption of AI that are applicable to state agencies — such as the increasing use of generative AI in Tier 1 and Tier 2 call centers. Another critical area is to show how AI can address the growing shortage of skilled workers by augmenting the skill sets of state agency employees to deliver more value to citizens while reducing budgets, allowing the state to do more with less. And the state’s Advisory Action Plan promotes “awareness of the successful implementation of AI technologies and guardrails within state government.”
Private AI: The AI Efficiency “Hack”
While policy and politics can be difficult, agreeing to speed up the benefits from AI technology need not be. Public sector technology advancements often require efficiency hacks that overcome resource challenges to enable innovation to deliver on its full potential.
Private AI is one efficiency strategy that can enable state governments to navigate the challenges of budgetary constraints, legacy technology investments, and the limited skill sets and dwindling populations of their technology workers. Already being pioneered in budget-minded global companies, this software-defined architecture can enable governments to build AI capabilities within a private cloud that takes full advantage of the virtualization stack anchoring their current data center infrastructures.
This AI implementation hack allows states launching generative AI solutions to start small and over time build out AI infrastructure one application at a time without breaking the budget.
Building on an existing foundation also minimizes the need for new hardware computing investments and curbs energy demands currently spiraling skyward among major cloud platform players. This is because virtualization enables AI capacity to be intelligently pooled and shared as needed, thus reducing both power requirements and costs when compared to other approaches.
Finally, because gaining and maintaining the confidence of a state’s citizens is imperative to AI success, the strong controls private AI delivers by its very nature ensure governments can keep their citizens’ data secure and private. With this approach, AI models are brought to agencies’ existing data, allowing organizations to reap the benefits of AI without compromising data privacy or control.
Immediate Benefits. New Efficiency.
The technical advances being pioneered in AI aren’t useful in the vacuum of a lab computer, they find benefit once they’re deployed on actual IT systems supporting modern digital services for state employees and citizens at large. Indeed, new innovations like private AI demonstrate the value in using today’s well-managed infrastructure in powerful new ways to reap immediate benefits for citizens and efficiency across government services, while keeping the human in the decision loop. AI is a tool that can intelligently assist humans, not replace them.
The nonpartisan analysts at MultiState note that “from a public policy standpoint, we’re solidly in the education phase of AI regulation. We’re all scrambling to get up to speed on this emerging technology.” They and others caution that while states consider their AI deployment model and strategies, they should also be wary of rushing ahead with legislation that might hinder AI innovation with premature attempts at regulation. The approach they advocate is: “Wait. Don’t Regulate.”
In a world where transformative and beneficial AI capabilities are being developed virtually overnight, that seems good advice indeed.