There’s a rush to adopt AI in most every American institution — including cities. Mayors across the country are rapidly integrating AI into public services. Los Angeles launched an AI zoning pilot for rebuilding homes lost to the Palisades Fires. Raleigh, North Carolina, is testing AI-enabled trash cans that can determine whether refuse is recyclable.
If this sounds more familiar than novel, you’re not wrong. The craze for municipal AI harks back to another era, from around 2015 to 2020, when “smart cities” were supposed to be the next big thing.
At the time, there was a notion that a smart city would integrate advanced technologies to fundamentally improve local governance. Plus, they’d have heated sidewalks, automate trash pick-up and offer apps to more efficiently deploy city services.
Those visions have largely fizzled. Sidewalk Labs, a former smart city subsidiary of Google’s parent Alphabet, had a $900 million initiative to transform a Toronto, Canada, neighborhood that was plagued with public furor over data privacy. The company blamed COVID-19 when it shut down the project in 2020, though that was two weeks before the city officials were to vote on whether to kill the effort themselves. Columbus, Ohio, cited the pandemic as well when it closed out its $50 million smart city initiative in 2021 — and similar privacy battles and bureaucratic hurdles were also at play.
To be sure, the idea of a smart city encapsulated more technologies than just AI. Yet its techno-utopian impulses seem to be enjoying a revival in the current buzz over AI-enhanced cities, for better or worse. Will this time actually be different?
“When I hear mayors and city leaders [...] talk about AI, I feel like I’m living in 2016 again,” said Ben Green, a University of Michigan policy professor who authored the book “The Smart Enough City.” “There are lots of lessons to be learned.”
On Thursday, Bloomberg Philanthropies announced it was adding 15 municipalities to its City Data Alliance, which launched in 2022 with a $60 million investment to provide tech support and training for local governments utilizing AI. Bloomberg Philanthropies connected DFD with mayors of four cities in this new cohort.
Their excitement was palpable. “Generative AI is on top of everyone’s mind, so if we could just move from an analog to digital world, that would be helpful for all of City Hall,” said Mayor Phillip Jones of Newport News, Virginia. He suggested that AI could analyze data from automotive collisions with deer to pinpoint where animal crossing signs would be most effective.
Mayor Matt Mahan of San Jose, California, said the city has become “the clearinghouse for use of AI in local government.” San Jose is piloting an AI system to optimize public transit routes using data collected from sensors on buses and traffic lights.
Mayor Quinton Lucas of Kansas City, Kansas, also brought up smart cities unprompted as a lesson in hubris. “Everyone was telling us, ‘Smart cities are all the rage, and we have the best smart city,’” he said. “Ten years on, you’re seeing that perhaps that system, while interesting, didn’t really change the way things work in the American city.”
Lucas pointed to the city’s smart kiosks that connect to interactive cellphone apps, which it began installing in 2016 for informational services. He says many are now broken. Lucas often walks by one sitting outside city hall, a cautionary relic that guides his thinking on using AI for more efficiently responding to 311 calls. “You can’t be too fixated on the technology for technology’s sake,” he told DFD.
James Anderson, who leads Bloomberg’s Government Innovation program, says the City Data Alliance learned lessons from the first generation of smart-city experiments. “The cutting edge of public sector technology adoption now focuses on real needs from real people first—and then figures out if tech can enable a better solution,” he told DFD. “The smart cities movement got that backwards.”
One sobering lesson from the smart-city era is that adding technology can lead to unintended outcomes. In 2017, San Diego made a big show of installing smart streetlights with sensors to collect data for alleviating traffic jams, only to shut them down a few years later due to public backlash over privacy. But they weren’t taken down — and in 2023, the city reactivated them for police surveillance.
In their enthusiasm, cities can also overlook the limitations or flaws of certain technologies, an issue from the smart-city era that is resurfacing in the age of AI. “They are always massively oversold by the tech companies,” Green said of systems marketed to cities. New York City released an Microsoft-powered AI chatbot last year to give people information on operating businesses. The chatbot ended up suggesting that bosses could pocket employees’ tips and that landlords could discriminate based on a prospective tenant’s income
One problem for cities, as they’re hit with a fresh wave of persuasive tech marketing, is that political cycles are short, and so are institutional memories. “The people who were around eight, 10 years ago and learned all of these lessons are now off doing different things,” said Green, “That’s pretty typical for turnover in city government.”
To their credit, the mayors that DFD interviewed seemed to be cognizant of what to avoid. Privacy and data analytics have matured as public issues in the past decade, and many of the mayors said they were implementing measures like anonymizing data to address the sorts of privacy concerns that bedeviled bygone smart cities. They professed to be clear-eyed about what they were really trying to do. “We’re not gonna be chasing cool stuff just to be chasing cool stuff,” said Mayor Tim Kelly of Chattanooga, Tennessee, who’s been thinking about using LLMs to make complicated city ordinances more intelligible. “I don’t think it’s too lofty to say that this has a potential to really increase, if not restore, faith in government.”
---
Autor(en)/Author(s): Aaron Mak; With help from Mohar Chatterjee and Steven Overly
Quelle/Source: Politico, 31.07.2025

