Using AI to build shared context for governing

Soon after I was elected mayor of Indianapolis in the 1990s, I cut a ribbon on a park my predecessor had started building. Expecting acclaim, I instead found complaints: Older neighbors were upset that the designers placed the basketball courts—and the nighttime noise that came with them—on their side of the park. The overworked parks staff left the event convinced that no good deed goes unpunished.

That ribbon-cutting went wrong for a reason local officials everywhere will recognize: The city and its residents were looking at the same place from different perspectives. Better engagement with people in their communities will always be central to how local governments understand residents’ needs and, in doing so, bridge that divide. Now, artificial intelligence can supercharge that work by serving as a context engine, connecting information that is often fragmented across people, places, and institutions to create a more deeply shared understanding on which residents and cities can act.

That shared context—that common view of local needs, services, conditions, and attitudes—enables residents and officials to explore the same scenarios as partners, from the potential to obtain new benefits to the possible design of a new bus stop. By creating a more unified perspective in this way, city leaders can do something critical at a moment of low trust in government: make residents feel more recognized, supported, and involved.

The most responsive governments recognize that residents have knowledge, networks, and lived experiences essential to solving complex problems. Here’s how effective public-sector leaders are starting to use AI to enrich decision-making with shared context, and why I believe cities should prioritize this work in the future.

Understanding (and acting on) resident needs.

The needs of vulnerable residents span agency lines—housing, health, benefits, workforce—with records fragmented across silos. City staff, no matter their specialty, can now use AI to build a clearer picture of residents’ circumstances and a shared view of the most effective interventions in their lives.

For starters, caseworkers can better understand the whole person by using AI to organize documentation, identify missing information, check eligibility, and more. This is already happening at various levels of government: For example, in New Jersey, leaders use AI to link residents’ records scattered across government systems—SNAP, Medicaid, and others—into a more complete view of their household. According to Chief Innovation Officer Dave Cole, this has helped the state identify about 100,000 families each year who were eligible for benefits but would not have received them otherwise. 

AI is also starting to give caseworkers more time and space to craft the best response for a struggling family, even when that response lies outside their area of expertise. In earlier research on accountability and discretion, my Harvard colleague Tony Yang and I argued that when cities use AI to support frontline workers, it creates flexibility for public servants and their community partners to exercise greater judgment. Now, cities such as Memphis are providing AI-enabled chatbots to community-based organizations so that case managers talking to a resident about something like domestic violence can also be in a position to answer questions about unrelated benefits. 

These early use cases reflect something critical: Cities can marshal AI to overcome information silos and get on the same page with residents about exactly where and how local government can show up in their lives.

Building collective insight into neighborhood conditions.

Just as frontline workers need a fuller picture of resident needs and the services that might help meet them, community representatives and city officials need shared context about neighborhood conditions. That requires easy access to spatially organized data so they can understand how traffic, housing, infrastructure, pollution, public health, safety, and resident concerns interact in a specific place. Going forward, mayors will use AI to provide that access.

For example, Detroit is exploring how to provide residents and city officials with natural-language spatial tools to answer questions such as, "How does the air around my child’s school compare to other schools?" or "Show me the traffic and industrial sources that aggravate her asthma." This approach will let residents and staff study how variables interact before committing to a remedy, so arguments over whether a hazard exists give way to shared evidence about where it does exist and how to tackle it. 

Boston already applies the same principle to everyday information through “OpenContext: AI for Boston’s Open Data.” The tool lets employees and residents use plain language to explore local datasets—no query skills or technical training required. 

With these approaches, cities are doing something my own experience in government proved to be absolutely essential: building shared context around neighborhood conditions before deciding how those neighborhoods might need to change.

Synthesizing resident feedback.

Shared context also depends on hearing residents directly. 

Public officials act neighborhood by neighborhood, but feedback arrives through many channels: comment portals, hearings, 311 calls, and petitions—often in volumes no staff can read end to end. Moving forward, effective governments will increasingly use AI to cluster comments, identify themes, summarize recurring concerns, detect outliers, and map each comment back to its source. That will make public input legible at scale and reduce the risk that officials hear only anecdotes or the loudest voices.

Already, many U.S. cities supplement traditional engagement with AI-driven sentiment mining of anonymized social-media data to enrich their sources of context and bring life to the voices of the often unrepresented. For instance, officials in Sugar Land, Texas, used this approach to spotlight the popularity of a new transit service to a skeptical city council. And cities like Bowling Green, Ky., now use AI to more systematically solicit, analyze, and act on input from wide swaths of their population. The next step will be for cities to use AI to map, in a tangible way, how individual government actions ladder back to feedback previously received from specific communities. In doing so, cities will prove that they’re not just listening, but responding. 

When used strategically, AI can help cities create a unified perspective that neither residents nor the government could build alone. That common view facilitates a shift from adversarial democracy to collaborative governance: Trust grows when residents know officials are seeing what they’re seeing. The end game? Repeated cycles of listening and follow-through that help build a shared information base and foster confidence that the government is approaching new and emerging challenges from the same vantage point as the people it serves.