This performance management tool has become a staple for state and local governments, but it’s time for a refresh. Recently, 10 cities came together at the Bloomberg Center for Cities at Harvard University to share how they can apply generative artificial intelligence (GenAI) tools to city performance. These early efforts in what we call a “StatGPT” model promise substantial improvements in local-government responsiveness.
StatGPT can transform city operations in three key ways:
Broader access to performance data: In the past, even the best stat programs depended heavily on strong personalities — think longtime policing leader Bill Bratton — and a small, specialized data team to pose the right questions and drive action. With GenAI tools, a much larger group of city employees can not only see results but dig into underlying causes, surfacing new questions and insights. Generative AI’s conversational interfaces, such as chatbots that use natural language, make it easier for more city workers, not just data specialists, to visualize and interpret data about city operations, 311, permitting, costs and time.
Routine analysis of regulations and contracts consumes valuable time from policy analysts and lawyers while often slowing responses. GenAI can unlock insights from open data, including a range of procurement, ordinance and regulatory postings, democratizing decision-making.
Enhanced accountability: New insights emerging from broader city hall and neighborhood participation will help establish a more collective understanding of performance. With a broader set of perspectives, officials can evaluate solutions and set new goals. Additionally, GenAI can be used to benchmark performance, making it easier to compare costs and response times across different jurisdictions.
With fast, personalized access to data and comparisons, supervisors and staff can see how their performance stacks up against peers. Mid-level supervisors, for example, can readily compare their unit’s performance to other teams doing similar work. Ultimately, these data tools will empower staff at all levels to ask more effective questions, make data-driven decisions and achieve better results.
Issue-focused — not just agency-focused — analysis: Traditional stat programs often confined analysis to departmental silos. Accessing cross-agency data poses a challenge, but GenAI-enabled agents make it easier to find connections across agency data walls, and in doing so will unlock how changes in one part of the government can influence outcomes elsewhere. GenAI tools can enable cross-functional problem-solving on issues such as public safety, housing or sanitation. Prompts, approved by the city and shared among employees, can bring in additional context, shape the analysis and deliver truly usable information.
The process of stat modernization should start not with data for its own sake but with practical questions: Who in city government is eager to improve performance? What information would help them do so? What GenAI-powered agents and bots can assist in identifying both the questions to ask and the answers? What technical training will be needed throughout the organization to be effective?
In partnership with community organizations and academic partners, Boston has explored how AI can be used to improve access to government reports and plans, as well as analyze open data. A prototype of AI search on boston.gov, developed by the Boston Digital Service, showed a quadrupled improvement in residents’ satisfaction compared to the traditional search experience.
With a rapidly changing landscape in government programs and regulations, community capacity to adapt to these changes is critical. GenAI can enable better experiences on existing platforms and also create new resources for community groups to learn, adapt and participate.
The future of performance management goes beyond GenAI and specific technologies. But for now GenAI can lead to a broader set of perspectives driving change, facilitating the faster diffusion of complex information with a lower barrier to acting on the knowledge, while reducing the difficulty of experimentation.
Governing’s opinion columns reflect the views of their authors and not necessarily those of Governing’s editors or management.
Related Articles
