Can AI Support Good Government? Some Pilot Programs Suggest Yes

Across the US, local and state governments are getting a feel for how artificial intelligence (AI) can help them in analytics, surveillance, operations, modeling, and more. A 2023 survey from Bloomberg Philanthropies found that while only 2% of local governments around the world reported using AI, many were exploring its potential. This blog will offer a snapshot of how governments in Washington State and elsewhere are using AI to accomplish a variety of ends.

Managing Traffic Control

According to the Environmental Protection Agency, each gallon of gasoline burned produces about 20 pounds of carbon dioxide, which can build up at stoplights and cause pollution hot spots.

Seattle has an ambitious goal of reducing greenhouse gas emissions 58% by 2030, and tackling congestion hot spots can play a role in this effort. It has become the first U.S.-based city to use Google’s Green Light technology to manage traffic signal timing. Green Light’s AI analyzes driving trends, smartphone maps, and model traffic patterns to make recommendations for traffic signal timing changes.

In 2024, Seattle identified and made changes to three major intersections in largely residential neighborhoods, resulting in improved traffic flow and reduced idling, and plans to extend the program in 2025 to an industrial area that houses several sports stadiums and transportation hubs.

Analyzing Law Enforcement Training

Thanks to a grant award from the U.S. Department of Justice, the Spokane County Sheriff’s Office (SCSO) will be using AI to conduct analysis of footage from body cameras worn by SCSO deputies.

Beginning in 2025 and through the next three years, the county will use TrustStat to analyze recorded interactions between SCSO deputies and the public, with a focus of identifying key behaviors and language related to de-escalation and use of force. The goal is to assess SCSO’s de-escalation training, which the agency will use to improve future offerings.

While SCSO is working with a private company for the current initiative, it has also worked with the Complex Social Interaction Lab at the Washington State University. This lab uses a combination of AI and human reviewers to analyze and report on bodycam footage of partner law enforcement agencies.

Improving Emergency Response

Multi Agency Communications Center (MACC) 911 in Grant County launched an AI system in May 2024 to free up dispatchers to address emergency calls, speed up the response time for non-emergency calls, and offer bilingual (English and Spanish) call support.

Because 911 call volumes spike during emergency events, call centers need to quickly determine which callers are experiencing actual emergencies and which are reporting on or seeking information about the event. AI-powered call diversion can filter out non-emergency calls before they reach a dispatcher, as well as aggregate geographic data of callers to help pinpoint the location generating the most calls.

When an individual calls MACC’s 911 line with a non-emergency, the AI asks the caller a series of questions and creates a report that is sent to a MACC dispatcher, who can then determine what type of follow-up is needed. If the report is unclear, dispatchers can consult the call transcript and recording. The system also allows callers with cell phones to share their location and stream live video to dispatchers.

According to the National Telecommunications and Information Administration, AI can improve the operations of 911 call centers, reduce the burden on dispatchers to manage non-emergency issues, and lead to improved response times. Importantly, AI integration into 911 call centers is happening at a time when many are finding it difficult to recruit and retain dispatchers.

Preventing Homelessness

Family Promise of Pierce County (Family Promise) implemented a new AI-based system that combines intake assessment, resource application, occupational service assistance, inter-system communications, and remote client support to help individuals quickly access a variety of social services available to them. Pierce County has a county-wide intake system, but bottlenecks develop when there are more people applying for services than social workers able to process incoming applications.

Applicants can complete Family Promise’s electronic intake assessment, which uses AI to generate a report detailing the individual’s needs and suggesting referral services, simplifying the process for the social worker. The system also stored frequently needed client paperwork, such as consent forms, and digital e-signatures.

Researches from UCLA have partnered with Los Angeles County to create an AI algorithm that can predict homelessness. By combing through county data, researchers found that people deemed to be at highest risk of homelessness tend to show up in emergency rooms and jails at high rates and have high usage of supportive services. One the AI was trained to spot risk factors, the county began using the system to identify and connect with at-risk individuals. The goal is to enroll such individuals in the county’s early intervention program, which offers temporary cash support and other services to keep people housed.

Modeling Foot Traffic

Local governments in Washington are also using AI-powered applications for economic development and tourism. In 2023, Grant County worked with Placer.ai to count the number of cell phones that entered and left the Grant County Fair and extrapolated an attendance count, including how many people visited multiple times.

The Snohomish County Tribune reports that Monroe and Snohomish are using Placer.ai to analyze foot traffic more broadly. Placer.ai foot traffic analysis can show how many people are at a given location, for how long, and how much money they spent while there. It may even be able to offer demographic information, such as average income and where visitors came from. So far, this analysis has resulted in:

• Snohomish increasing police staffing at the city’s seven annual big events.
• Targeted marketing campaigns to areas that generated the most visitors to Monroe the prior year.
• Data on the effectiveness of activities conducted by recipients of Snohomish County lodging tax funds.

AI-powered datasets can combine historical foot traffic, social media trends, web searches, and weather patterns to predict when and why tourists will visit a local destination, allowing agencies to anticipate fluctuations in foot traffic, plan for peak hours, and simulate the potential impact of increased tourism on local resources.

Detecting Wildfire

In 2023, the Washington State Department of Natural Resources (DNR) announced that it had installed AI-enhanced cameras in areas most at risk for wildfires. The Pano AI stations combine 360-degree, ultra-high-definition cameras with AI-powered software to search for signs of fire within a 15-mile radius. If a station spots smoke, it sends an immediate notice to local incident command centers pinpointing the exact location and including images with topographical maps. The Pano AI station can also identify the nearest water source and predict how wind patterns may influence the fire.

Advanced notice from a Pano AI station helped firefighters put out the Crater Creek Fire in 2023 before it grew. DNR eventually hopes to grow a statewide network of cameras (there are 21 currently deployed) in partnership with local fire depts and other partners, such as Puget Sound Energy, which installed Pano AI stations in remote parts of its service area as part of a wildfire mitigation plan (2024).

Conclusion

While there is considerable hype around AI adoption, a public agency should conduct the same thoughtful analysis of the potential risks and rewards that it would before adopting any new technology.

MRSC writers have been addressing topics local governments should think about before employing AI, such as public records considerations, and privacy, transparency, and cybersecurity concerns. Readers may also want to check out the following sample AI use guidelines from MRSC’s sample document library:

• Generative AI Guidelines – San Jose, CA (2025)
• Generative Artificial Intelligence Policy – Seattle (2023)
• Generative Artificial Intelligence Use Policy – Kirkland (2024)
• Interim Guidelines for Using Generative AI – Boston, MA (2023)