According to Code for America, artificial intelligence is increasingly reshaping the delivery of public services across the United States, but adoption remains uneven and difficult to track. While many state governments are experimenting with AI through low-risk pilot programmes, they often lack clear frameworks to evaluate its real-world impact and effectiveness.
The report outlines a staged approach to AI adoption, beginning with a readiness phase focused on building leadership capacity, workforce skills, and technical infrastructure. This is followed by piloting, where governments test AI tools through controlled experiments, and then implementation, where systems are integrated into everyday operations such as fraud detection, document processing, research support, and citizen-facing digital assistants.
Despite growing experimentation, most states have yet to fully transition to mature, operational systems where outcomes can be consistently measured. Some states, including Utah, New Jersey, Pennsylvania, North Carolina, Maryland, Texas, and Vermont, are progressing in developing the institutional capacity needed to manage AI as a long-term public asset, while others remain at earlier stages of adoption and preparedness.
A central challenge identified in the report is the difficulty of measuring outcomes. Although AI is expected to improve efficiency and reduce costs, initial deployment often increases workloads for public employees before benefits are realised. At the same time, evaluation frameworks are still underdeveloped, limiting governments’ ability to clearly assess performance improvements.
Amanda Renteria emphasizes that the real opportunity lies not just in adopting AI, but in ensuring it is deployed in a human-centred way with measurable public outcomes. The report concludes that states capable of aligning technological innovation with tangible community impact will move beyond experimentation and help shape the future of AI-driven public services.
