When it comes to quality of life, New Yorkers have a lot to complain about: public urination, intoxication, trash accumulation, and homeless people in the transit system. As of this week, year-to-date quality-of-life-related calls to 911 have increased by 8 percent, while transit-related calls have also risen. Mayor Zohran Mamdani has said that he wants a government that works for the people. That’s a big challenge in a city where only 34 percent rated quality of life as “excellent or good” in 2025, compared with 51 percent in 2017.
If Mamdani is serious about improving the city’s quality of life, small changes around the edges will help. But as I argue in a new Manhattan Institute report, the best tool for addressing the largest and most blatant problems is artificial intelligence.
Finally, a reason to check your email.
Sign up for our free newsletter today.
The provision of city services is a resource-allocation problem. How should an agency deploy limited resources most effectively?
New York City has a history of using data-driven governance to answer these questions. The NYPD’s CompStat initiative, for example, succeeded because it made the work of policing predictive rather than a guessing game. The same principle should inform quality-of-life problems.
That’s where AI comes in. By utilizing AI, the city can preempt problems and properly distribute resources in the most effective way.
AI differs from previous data-driven solutions in two important ways. The first is multimodal processing. Traditional analytics pipelines require nicely formatted rows and columns. They cannot process a 311 complaint written in English, a photo of a cracked sidewalk, a noise complaint recorded as audio, or a security camera feed. AI can.
The second AI innovation is speed at scale. The bottleneck in data initiatives has always been the capacity to synthesize information quickly enough to act on it. Having to take that extra, manual step in between something occurring and formatting it as usable data is valuable time lost. Systems incorporating AI need many fewer of these steps.
How, concretely, can AI help with quality-of-life issues? Take drag racing, a behavior that has plagued city neighborhoods. Currently, the problem is akin to whack-a-mole: reckless driving in one area occurs; someone complains; maybe the police show up; the next night, the drivers reappear a few blocks away.
The data to solve this problem already exist. These include records of 311 complaints, speed-camera violations, traffic volume, street geometry, NYPD incident narratives, patrol schedules, and so on. But these data aren’t synthesized because it would be nearly impossible to plug these streams into a single output.
Nearly impossible, that is, without AI. A vision model can process the camera footage to flag the telltale mark of a drag racer’s burnout. A text model can parse 10,000 complaints and tag them by behavior, time, and location. Once those variables exist, regular regression can do the forecasting to figure out where reckless driving is most likely to occur. Government agencies can then make a data-driven decision about where to put the patrol on a Saturday night, and where the next speed camera should go.
Some may worry that such a comprehensive system would threaten our freedoms. But civil-liberties concerns should not impede us from taking already existing data and putting them to better use. The city doesn’t need social credit scores or GPS tracking of every citizen. It simply needs to find a way to take preexisting data from our streets and use this information toward a better end.
But before any of this can happen, the city must fix its data. NYC Open Data, which publishes much of the data necessary for this project, has 2,412 datasets and more than 6 billion rows. Yet 437 of these datasets remain unautomated, meaning they must be manually updated to remain up to date. It took the Office of Technology and Innovation 12 years to automate 435 sets. AI would be the best tool for this task.
Adding AI doesn’t mean removing humans from the loop. CompStat worked because commanders had to answer for the numbers every week. The AI equivalent would replicate the same accountability with better data. The technology supplies the foresight, but only the mayor and city agencies can enforce the consequences.
New York has the technology in place to make life better and safer. The question is whether we will implement these tools to capitalize on the benefits they offer.