In the city of Raleigh, Mark Wittenburg is keeping a close eye on traffic—with a little help from artificial intelligence.
Wittenburg, the North Carolina capital’s CIO, oversees efforts to make AI part of its operations. Through a pilot project with Nvidia and GIS mapping software maker Esri, Raleigh is using AI cameras and vision tools to study traffic patterns and counts at several intersections. As a Vision Zero city, it aims to eliminate traffic fatalities. AI offers a way forward.
“We know when an accident happens, but we don’t know what are the near misses,” says Wittenburg, who leads an innovation team of about 10 people. “What we’re trying to do right now is look at the technology and see if we can train the AI to start looking for … those near misses between vehicular traffic and pedestrians. And then how can we make adjustments to make those intersections safer?”
To that end, Wittenburg and his team have big plans. “We’re trying to get the funding to be able to take this pilot from three or four intersections up to every intersection in the city.”
Raleigh is just one of many U.S. cities that are using AI to make themselves smarter, safer, and greener, improving the lives of citizens along the way. Although they arguably lag some of their peers in elsewhere the world, American municipalities have no shortage of potential.
At the city level, AI is gaining traction. In a recent global survey, Bloomberg Philanthropies found that even though just 2% of cities are implementing generative AI, almost 70% are exploring or testing it.
AI holds plenty of upside for cities. Junfeng Jiao, an associate professor in the Community and Regional Planning Program at the University of Texas at Austin, sees three main benefits.
First, AI will help cities operate much more efficiently, says Jiao, whose research focuses on designing, developing, and deploying ethical AI in cities to better serve the community and its residents. Transportation, housing, water, electricity, information, and fire and police services are some areas that could see improvements.
Second, if U.S. cities capitalize on AI, it will yield new companies and industries, says Jiao, who is the director of the Texas Smart Cities project. “This is a huge employment opportunity for us, and also [to] export our technology to other countries.”
Third, life will be more enjoyable for city residents. As the Raleigh project suggests, for instance, using AI to help manage traffic could reduce accidents. It can also combat congestion, which means less gridlock and lower carbon emissions.
Cities have been using AI and advanced analytics for years, observes Leigh Sheldon, a partner at consulting firm Guidehouse. They have, for example, deployed machine learning to build predictive models, says Sheldon, who leads Guidehouse’s data and analytics solutions for clients that include state and local governments. “What we are now seeing, though, is the emergence of generative AI, which takes the maturity to a whole other level.”
For cities, generative AI is both exciting and overwhelming, Sheldon says. “It’s exciting because we all like to ideate through the art of the possible, but there’s also a lot of component pieces that need to be in place in order to most effectively leverage generative AI.”
Elizabeth Crowe, director of the City of Cleveland’s Office of Urban Analytics and Innovation (Urban AI), says most of the cities she’s talked to are enthusiastic about generative AI. “‘Where can we gain efficiencies? Where can we optimize our work? How can we better serve our residents?’—because that’s really what we do at the end of the day. But I think we’re truly in an experiment-and-learn mode.”
There’s also a lot of work to do. “We have the technology capacity and the resources, and we have the human power to be the [leader] of the world in terms of using AI in cities,” Jiao says. “But unfortunately, we are not there yet.”
Singapore, Seoul, and Dubai are way ahead when it comes to using AI to manage, design, plan, and develop cities, Jiao says. Japan also has many good examples, he adds.
In the U.S., AI is widely used at the city level, Jiao says, but the projects tend to be one-offs. There might be smart cameras in Austin, a smart water system in New York, and a digital twin for Dallas. “But you don’t have all of them in one city.”
America lags behind for two reasons, Jiao says. Its cities must build residents’ trust in AI, and they need funding. “Most of the AI applications in cities will require significant investment.”
For Raleigh and other cities, part of the motivation for using AI is to provide the level of service that people enjoy from tech businesses in their daily lives, Wittenburg explains. “Our community is going to expect the same thing from us as they see from these other companies. They’re saying, ‘Why can’t we have that at the city level?’”
For example, the number one call to Raleigh’s Solid Waste Services department is about missed household trash cans, Wittenburg says. Raleigh could use AI cameras to spot which cans aren’t at the curb on pickup day, and then notify their owners that the truck is coming. “That would be great customer service, improve efficiency, and raise the community positivity around things.”
In the meantime, Raleigh has found other applications for AI. The city is using it to forecast breakdowns in its water and stormwater systems, based on factors such as age and condition of infrastructure, soil materials, and previous failures. Wittenburg’s team feeds all of those variables into an AI model that says, “‘Okay, these are the areas where we predict it might fail, and these are the areas where we think we need to make investments into our infrastructure,’” he notes.
Raleigh is deploying AI to address climate change, too. With partners including not-for-profit MITRE Corp., the city created a digital twin of itself paired with AI-powered microclimate modeling. The goal: tools and data-driven evidence that show how urbanization can be more adaptable to climate change. That could mean designing for environmental sustainability and resilience, creating more comfortable public spaces that take “real-feel” temperatures into account, or identifying and mitigating risks like heat islands.
The digital twin project is a finalist in the 2024 Smart Cities North America Awards.
In Cleveland, Crowe is building out Urban AI’s data analytics capabilities and laying the groundwork for use of advanced AI.
When she joined the city in late 2022, her office hired a handful of data staff. The team has been setting up a central data warehouse, along with analytics and reporting for checking data quality, Crowe says. “And then starting to get us ready for whatever the next step is going to be for cities. A lot of cities are still trying to figure that out right now.”
Urban AI’s first big use case is what it calls parcel analytics—integrating the scattered data being collected about pieces of land and the properties on them.
“If I think about a specific part of the city, are police, fire, and EMS going there?” Crowe asks. Along the same lines, are properties being cited for code violations, or is a property owner a tax delinquent? “All of that data lives in different systems across the city,” Crowe says. “And so step one is just to start with a data management exercise.”
For cities, the elephant in the room is generative AI, whose potential uses span urban design, improved transportation, and chatbots trained to answer citizens’ questions. Looking ahead, Raleigh’s Wittenburg thinks it’s the biggest challenge. “This is a freight train that’s coming,” he says. “How do we equip our organization to be able to take on this new technology?”
That raises policy and other questions. “How do we tackle, perhaps, bias in our data that’s going to be feeding these models?” Wittenburg asks. “And then how do we verify that these models and the information that [they’re] providing our citizens is accurate?”
With that in mind, the City of San Jose has established the GovAI Coalition, a partnership of some 150 local, county, and state governments whose mission is to promote responsible and purposeful AI in the public sector.
“Very few cities are using [generative AI] at its full potential,” says Raveen Rao, a partner in the state and local government practice at Guidehouse. That’s because of considerations like safety and ensuring that they’re acting in good faith with citizens and employees, he adds. “So I think a lot of cities are just sort of wrapping their arms around it, although we do see a few that are doing some pilots.”
For the City of Cleveland’s workforce, the first phase of generative AI might include using a chatbot to help write a memo, Crowe says. Employees could follow guidance like that issued by Boston, she explains. “That’s a low-risk, low-cost exercise where we can gain efficiencies in our day-to-day operations.”
The next phase: finding data that’s suitable for an AI or machine learning model. For example, Cleveland is looking at how to optimize its road-quality assessment. Historically, the city has had an engineering firm survey all of its streets by hand—a process that takes five years.
Enter AI. “There are companies on the market now that can drive your entire city; they can scan or take an image of the roads through the entire city, and then use an AI model to give you back a road-quality rating,” Crowe says. “That’s a little higher-risk, because the model could give you something weird,” she admits. “But you’re collecting data that’s fit for purpose for that model, which to me is the next phase for cities—these tangible and specific use cases.”
On that note, Crowe encourages cities that are thinking about using AI to start by finding such a use. “You can succeed at narrow and specific in a reasonable amount of time; you can learn a lot,” she says. “And then go on to the next, slightly bigger use case.”
When it comes to AI, what needs to happen at the city and regional level for the U.S. to catch up with some other parts of the world? It calls for a change of mindset, UT-Austin’s Jiao argues. For starters, cities can support pilot projects with startups and established companies to test AI technology.
As Jiao sees it, cities are a huge untapped market for AI. “We want to lead this,” he says. “We don’t want some other companies to come to the U.S. and say, ‘Adopt our technology because we already tested it, we already proved it.’”
Asked what the future holds for cities and AI, Guidehouse’s Rao zeroes in on the productivity and efficiency gains the technology can enable. “The ability for cities and states and counties to do more will be just amazing,” he says. “I think it’s gonna just blow our minds.”
