8 ways AI is changing the way local governments do business
Local governments are turning more frequently, and in an increasingly broad variety of ways, to artificial intelligence. These eight examples show just how great an impact AI is making in the lives of people around the world.
Обобщение
Local governments have always needed to balance public service with financial prudence. The last few years have added a pandemic, a global economic downturn, and an increasingly pressing need to emphasize sustainability to the already daunting list of considerations for the public servants who keep our city and county governments running.
Greater challenges demand more effective and efficient responses. That’s why local governments are turning more frequently, and in an increasingly broad variety of ways, to AI. These eight examples show just how great an impact AI is making in the lives of people around the world.
Aarhus, Denmark
Environmental sustainability plays a key role in city planning, zoning, and development. But how can municipalities assess the environmental impact of the companies with whom they contract for goods and services? Aarhus, Denmark’s second-largest city, answered that question with the help of Konsido ApS, a local company that has made a global impact with its approach to automating accounting processes.$ $
For Aarhus, Konsido studied every invoice issued by the city in the past several years. It used AI-driven historical analysis to develop a classification system of more than 1,000 categories, then drew on a huge trove of contemporary data to create estimates of the carbon emissions generated by the activities associated with each category. Aarhus spends roughly DKK 200 billion with authorized buyers each year, and its partnership with Konsido gives it a much better idea of how to steer public expenditures toward more sustainable options, and how to set a course for more sustainable civic development into the future.
Phoenix, Arizona
The Phoenix metro area is one of the fastest-growing in the US, and with more people comes more traffic. That’s bad for commuter times, emergency response, and public safety: Phoenix has one of the nation’s highest rates of pedestrian fatalities.
None of this is new—cities have long sent human observers to monitor problematic roads and intersections in an effort to better understand traffic patterns and plan strategic responses to traffic-related issues. But human error often skews those measurements, and neither road sensors nor mobile data provides the kind of granularity needed to make informed decisions about traffic management.
To improve on those approaches, the City of Phoenix Street Transportation Department recently partnered with Argos Vision, a startup launched in 2022 by two Arizona State researchers at ASU’s School of Computing and Augmented Intelligence. Argos provided a neural network so well programmed and tailored to the task of traffic recognition that it is able to deliver remarkably detailed information about specific roads and intersections without the cumbersome overhead imposed by earlier projects of its type. $ $
Phoenix provided traffic. Lots of it, especially near City Hall and the ASU campus. The ongoing project seeks to reveal the rate and nature of both traffic congestion and conflicts between vehicle and pedestrian traffic.
Vernon, British Columbia
Wildfires have always plagued the Pacific Northwest, and in recent years they have become more frequent, more intense, and more destructive. Teaming with the University of British Columbia, the town of Vernon, BC sought to augment the information it gathered through fire patrols and reports from residents by deploying 100 small sensors throughout the city and in part of the adjoining Ellison Provincial Park.
SenseNet, as the project has come to be called, covers up to 100 hectares per sensor, depending on its placement. Though the sensors are tiny, they are equipped with AI technology that transforms them from passive data collectors into powerful analysis tools sensitive enough to distinguish between a campfire and a nascent wildfire. $ $
That much has been proven in controlled settings. The Vernon pilot, which runs through August 2024, is SenseNet’s first opportunity to prove its value in real life. Public funding and the UBC’s central role in the project limit the costs incurred by Vernon, but a successful debut by SenseNet could lead to wider deployments and economies of scale that make it readily available to municipalities and private landowners throughout wildfire country.
Christchurch, New Zealand
New Zealand’s second-largest metro area and the largest city on its South Island, Christchurch also has a recent history with wildfires: more than 2,000 hectares burned in 2017, forcing more than a thousand Christchurch residents out of their homes. Like Vernon, Christchurch turned to sensors in its vulnerable Waitākiri/Bottle Lake Forest Park. Christchurch’s project, though, is in some respects a bit more ambitious. $ $
The park’s entire 800 hectares are monitored by five solar-powered sensors designed to take readings of ground temperature and precipitation, which provide the system’s AI algorithms the data they need to assess the likelihood of wildfires and to alert local authorities to issue appropriate warnings. The monitors also assess air quality, allowing them to warn of high pollen counts and the presence of airborne particulates associated with fire. The system’s predictive modeling puts Fire and Emergency New Zealand in an unaccustomed but welcome position: fire-safety crews are now able to pre-emptively address wildfire conditions before fires break out.
Memphis, Tennessee
Potholes are more than a nuisance for motorists: if not spotted and fixed early on, they can wreak havoc on municipal infrastructure budgets, too. That’s why the City of Memphis turned to SpringML to see if AI could help interpret camera data to identify potholes as they emerged.
The project combined video from city buses and code enforcement vehicles with location data from calls to municipal service lines, most of them 311 calls, reporting potholes throughout Memphis. With training, the AI models that interpreted camera images were able to identify potholes with greater than 90% accuracy.$ $
To build historical and trend analysis into their models, the SpringML team worked with Memphis officials to add paving and traffic information from the city’s GIS, along with survey data and property records. This additional layer of analysis allowed the project to identify patterns of pothole development, greatly increasing its ability to focus resources on streets and neighborhoods that most needed them.
As a result, Memphis is able to respond more quickly than ever before to small potholes before they become big ones. The city’s motorists aren’t the only ones who benefit: Memphis has saved an estimated $20,000 in claims resulting from potholes that go unfilled for too long.
Owen Sound, Ontario
North America’s roads and bridges need serious attention, and cities know that improvements to their transportation infrastructure can pay immediate and lasting benefits. But there’s a catch: it costs a great deal to repair, reconstruct, and resurface streets and roads, leaving many municipalities unable to make the kind of investment needed to reap the massive rewards that follow those sorts of projects.
To make better use of its asphalt-resurfacing budget, the city of Owen Sound, Ontario implemented IrisGO, an AI-powered technology that accurately documents the condition of each segment of road throughout the town. This approach represented a significant advance in at least three respects: it was more efficient than the manual examinations on which Owen Sound, like most cities, used to rely; it collected a greater array of information on road conditions than previous methods; and it allowed Owen Sound’s Public Works and Engineering staff to assess the conditions of the city’s roads on a consistent, objective scale.$ $
The resulting Pavement Condition Index gave the city the information it needed to draft a ten-year resurfacing plan, and to make the best possible use of its resurfacing budget. It also allows Owen Sound to confirm the success of its recent resurfacing efforts: as it happens, the number of road segments with a Pavement Condition Index of 65 or better has risen 11% since 2019.
Milwaukee County, Wisconsin
Inequality is a pernicious aspect of our housing system, and municipalities invest a good deal of time and effort identifying inequality in everything from leasing and mortgage lending to foreclosures. Milwaukee County sets a high standard for housing equality, and has gone so far as to declare racism a public health crisis in a 2020 ordinance.
But the best-intentioned policies amount to little if they do not produce meaningful, measurable results. In that spirit, Milwaukee County formed Team MKE Equity and participated in the 2022 SAS Hackathon. Its project: an AI-assisted analysis of its own foreclosure policies, regulations, and ordinances against the historical pattern of foreclosures throughout the county.$ $
Using historical foreclosure data and a wide range of demographic and geographic information, the team was able to analyze the effects of certain types of policies—and even certain language used in those policies—to real changes in the equitability with which home foreclosures were enforced throughout the county. The result is a much clearer idea of what works and what doesn’t, and a better basis for enacting even more effective policy.
Mendoza, Argentina
With a metro population of over a million, Mendoza is Argentina’s fourth-largest urban area, and its most sprawling. Since the end of WWII, the population of Greater Mendoza has grown fivefold, challenging its essential services to keep up.
Chief among those services is waste management. Its large area and relatively low population density combine to create plenty of open space in Mendoza, and illegal dumping had become a problem. To readily identify areas serving as open-air waste dumps, Mendoza’s Department of Environment and Sustainable Development worked with the BYB Foundation of Argentina to develop a system that regularly monitors the city for illegal dumping.$ $
Gathering aerial images of Mendoza wasn’t a problem. Scanning those images and correctly identifying sites of illegal open-air trash disposal was another thing altogether. With the help of AI-powered analytical tools developed specifically for Mendoza, the city was able to automatically process the results of high-resolution aerial imaging, and to send those results to the appropriate authorities.
More AI use cases
For more examples of how local governments are utilizing artificial intelligence to improve services, streamline processes and reduce costs, check out our growing collection below.