OCTOBER 18, 2023

State of Cities:
Generative AI in Local Governments
Contents

1. Executive summary 4. Opportunities and barriers

2. Mayors’ interest in generative AI 5. Areas of collaboration

3. Cities’ use of generative AI 6. Methodology

2
1 Executive summary

3
Executive Summary

Generative AI (Gen AI) MAYORS’ INTEREST IN GENERATIVE AI

OPPORTUNITIES AND
promises to revolutionize the way BARRIERS
cities work and, with strategic
• 96% of surveyed mayors expressed interest in using Gen AI. • Opportunities for use of generative
implementation, has the potential AI among cities include improving
• Their biggest questions center around the technology’s
to vastly improve local implementation, impact on city services and efficiency, citizen engagement (81%),
governments’ efficiency, its and its ethical, legal, and social implications. enhancing data-driven policymaking
capacity to innovate, and, most (76%), and optimizing service
delivery (74%) and administrative
importantly, its efforts to improve processes (70%).
residents’ lives. Getting there, CITIES’ USE OF GENERATIVE AI
• Cities reported multiple barriers to
however, necessitates a better Gen AI adoption, with 74% citing
• While only 2% of surveyed cities are actively implementing
understanding of how this new insufficient technical expertise, 72%
Gen AI, 69% are exploring or testing the technology.
technology is already being put indicating a lack of awareness, and
• 22% of surveyed cities have designated a Gen AI lead, 13% 70% referencing budget constraints.
to work, where it presents the have developed policies and guidelines around its use, and
best opportunities, and what 11% have provided Gen AI training to staff.
pitfalls could get in the way of • Cities see the most potential in leveraging Gen AI to AREAS OF
progress. This research, which address traffic and transportation (34%), infrastructure COLLABORORATION
included a two-survey process (24%), public safety (21%), environment and climate (21%),
and education (18%). • Cities expressed strong willingness
including 100 Mayors and city
• Most cities reported exploring the use of Gen AI for data to collaborate and share expertise
staff from around the world, in using Gen AI to tackle urban
analysis (58%), citizen-service assistance (53%), and
addresses these questions and drafting memos, documents & reports (47%). challenges, in advancing Gen AI
key trends in today’s evolving • Four in five cities reported that security and privacy (81%), knowledge, and in developing
Gen AI landscape, including: and accountability & transparency (79%) are the key ethical best practices and policies for
principles guiding their exploration and use of generative AI. responsible use of Gen AI.
4
2 Mayors’ interest in generative AI

5
80 mayors across the globe participated in the research
Adama, Ethiopia Knoxville, Tennessee, USA Rourkela, India

Allentown, Pennsylvania, USA Kumasi, Ghana Sacramento, California, USA

Amherst, New York, USA Lancaster, Pennsylvania, USA San Bernardino, California, USA

Banjul, The Gambia Lansing, Michigan, USA San Francisco, California, USA

Baton Rouge, Louisiana, USA Lincoln, Nebraska, USA San Pedro Garza García,
Monterrey, México
Butuan City, Philippines Liverpool, England
Sandy Springs, Georgia, USA
Charleston, South Carolina, USA Maipú, Región Metropolitana, Chile
Santa Fe, New Mexico, USA
Chattanooga, Tennessee, USA Masaka, Uganda
Sarajevo, Bosnia and Herzegovina
Columbia, South Carolina, USA Missoula, Montana, USA
Scottsdale, Arizona, USA
Dubuque, Iowa, USA Mogi das Cruzes, São Paulo, Brasil
Scranton, Pennsylvania, USA
Durham, North Carolina, USA Moncton, New Brunswick, Canada
Sintra, Portugal
Elizabeth, New Jersey, USA Monterrey, Nuevo León, México
Skopje, Macedonia
Fargo, North Dakota, USA Montevideo, Uruguay
South Yorkshire, United Kingdom
Fort Collins, Colorado, USA Nansana Municipality, Uganda
St. Louis, Missouri, USA
Freetown, Sierra Leone New Bedford, Massachusetts, USA
St. Petersburg, Florida USA
Gezer, Israel New Orleans, Louisiana, USA
Stamford, Connecticut, USA
Glasgow, Scotland Oklahoma City, Oklahoma, USA
Tacoma, Washington, USA
Greater Manchester, United Kingdom Paterson, New Jersey, USA
The Hague, Netherlands
Hampton, Virginia, USA Providence, Rhode Island, USA
Tirana, Albania
Helsinki, Finland Quelimane, Mozambique
Torino, Italy
Hermosillo, Sonora, México Quillota, Valparaíso, Chile
Tulsa, Oklahoma, USA
Highland Park, Illinois, USA Raleigh, North Carolina, USA
Turku, Finland
Huntington, West Virginia, USA Regina, Saskatchewan, Canada
Vancouver, Washington, USA
Jackson, Mississippi, USA Renca, Chile
West Palm Beach, Florida, USA
Kansas City, Kansas, USA Reykjavik, Iceland West Sacramento, California, USA
Kitchener, Ontario, Canada Rochester, Minnesota, USA White Plains, New York, USA
Youngstown, Ohio, USA

6
78% of Mayors said they are interested or extremely interested in using generative AI

39% 39%

(I am) interested in
how AI might help us
predict where house
fires might occur,
18% how we might lower
healthcare claims,
and predict street
violence…
2% 1%
Extremely Interested Somewhat Somewhat not Not interested
interested interested interested

7
Key questions from Mayors about generative AI in cities centered around 8its implementation,
impact on city services and efficiency, and ethical, legal, and social implications

IMPLEMENTATION AND ADOPTION OF WORKFORCE TRANSFORMATION AND JOB IMPACTS

GENERATIVE AI IN CITIES

• Exploring opportunities and challenges in adopting generative AI • Assessing the impact of AI on current jobs and future job growth
• Identifying the sectors and services where AI can be applied • Addressing concerns about unemployment, displacement, and skill requirements
• Ensuring data privacy and addressing risks associated with implementation • Redefining roles to focus on more meaningful aspects of work with AI support
• Building governance and policies for responsible use of generative AI • Balancing labour group resistance with the potential benefits of automation

ETHICAL, LEGAL, AND SOCIAL IMPLICATIONS OF

IMPACT ON CITY SERVICES AND EFFICIENCY
GENERATIVE AI IN CITIES

• Enhancing communication and effectiveness of city services through AI • Examining legal frameworks for using generative AI in government settings
• Improving customer service while ensuring information security • Addressing ethical considerations, privacy concerns, and bias mitigation
• Optimizing resource allocation and decision-making through data analysis • Promoting transparency, accountability, and democratic governance in AI use
• Creating efficiencies in delivering city services and improving quality of life • Evaluating the socio-economic implications of generative AI adoption

Note: Results from thematic analysis of narratives shared by 82 mayors across the globe 8
3 Cities’ use of generative AI

9
53 of responding cities provided further details on their generative AI plans and priorities

Adama, Ethiopia Montevideo, Uruguay

Allentown, Pennsylvania, USA Nansana Municipality, Uganda
Amherst, New York, USA New Bedford, Massachusetts, USA
Banjul, The Gambia Paterson, New Jersey, USA
Butuan City, Philippines Providence, Rhode Island, USA
Charleston, South Carolina, USA Quelimane, Mozambique
Chattanooga, Tennessee, USA Raleigh, North Carolina, USA
Dubuque, Iowa, USA Regina, Saskatchewan, Canada
Fargo, North Dakota, USA Reykjavik, Finland
Freetown, Sierra Leone San Bernardino, California, USA
Glasgow, Scotland San Francisco, California, USA
Greater Manchester, United Kingdom San Pedro Garza García, Monterrey, México
Helsinki, Finland Santa Fe, New Mexico, USA
Hermosillo, Sonora, México Sarajevo, Bosnia and Herzegovina
Huntington, West Virginia, USA Scranton, Pennsylvania, USA
Kansas City, Kansas, USA South Yorkshire, United Kingdom
Kitchener, Ontario, Canada St. Louis, Missouri, USA
Knoxville, Tennessee, USA
St. Petersburg, Florida USA
Kumasi, Ghana
Stamford, Connecticut, USA
Lancaster, Pennsylvania, USA
The Hague, Netherlands
Lansing, Michigan, USA
Tirana, Albania
Maipú, Región Metropolitana, Chile
Torino, Italy
Masaka, Uganda
Tulsa, Oklahoma, USA
Missoula, Montana, USA
Turku, Finalnd
Mogi das Cruzes, São Paulo, Brasil
West Sacramento, California, USA
Moncton, New Brunswick, Canada
White Plains, New York, USA
Monterrey, Nuevo León, México

10
City halls are at different stages of leveraging AI, with only 2% actively implementing
the technology, while 69% are still exploring or testing

60%

we are just
beginning our
28%
exploration on
appropriate uses
9% of (generative) AI

2%
Not using Exploring Testing Implementing

• Not using indicates that the city has not yet started exploring or using Generative AI
• Exploring implies that the city is in the initial phase of researching and learning about Generative AI
• Testing implies that the city is actively experimenting with Generative AI​
• Implementing implies that the city has integrated Generative AI into some or all of its operations

*Note: total percentages do not equal 100% due to accuracy in rounding 11

71% of cities are exploring, testing or implementing the use of generative AI,
yet most cities have not developed their capabilities and policies

22% 13% 11%

cities have designated cities have set policies/ cities have provided
a generative AI lead1 guidelines around training to their staff on
generative AI use generative AI1

1 Based on responses from 50 out of the 53 cities 12

 13
34% of cities reported being interested in leveraging generative AI for traffic and
transportation followed by infrastructure, public safety, environment and climate, and
education sectors

Traffic and transportation 34% (We have) explored

generative AI
Infrastructure 24%
ancillary to other
Public safety 21% projects. Examples
Environment and climate 21% include "smart"
technologies such
Education 18%
as parking,
Tourism 16% transportation,
Water and sanitation 13% SCADA for water
and water resource
Emergency response 13%
recovery and
Energy and electricity 11% public safety.
Healthcare 11%

13
 14 analysis
Cities most commonly reported exploring the use of generative AI for data
(58%), citizen service assistance (53%) and drafting memos, documents and
reports (47%)

Data analysis 58%

(We are) leveraging

Citizen service assistance 53% generative AI to help
in conveying
Draft memos, documents, reports 47%
technical information
to residents in a
way they can
Administrative tasks 45%
understand.

Coding and programming tasks 34%

14
A large majority of cities reported that security and privacy (81%), and15accountability
& transparency (79%) are the key ethical principles which guide their exploration and
use of generative AI

Security and privacy 81%

There are
Accountability and transparency 79% concerns
regarding potential
Reposible use 58% impacts on our
cyber security
Empowerment and inclusion 43% status which is
something of great
Impact potential 32% importance to us.

Reliability 32%

15
4 Opportunities and barriers

16
Opportunities for use of generative AI among cities include improving citizen
engagement (81%), enhancing data-driven policymaking (76%), optimizing service
delivery (74%) and administrative processes (70%)

Improving citizen engagement 81%

Enhancing data-driven policy making 76%

Improving our
Optimizing service delivery 74% service delivery
and increasing
Streamlining administrative processes 70% staff efficiency is
our top priority.
Streamlining communication 63%

Optimizing resource allocation 57%

Supporting innovation and research 52%

17
Among cities not using generative AI, opportunities for desired future use include
improving citizen engagement (93%), enhancing data-driven policymaking (80%)
and optimizing service delivery and administrative processes (73%)

Improving citizen engagement 93%

Enhancing data-driven policy making 80%

Improving our
Optimizing service delivery 73% service delivery
and increasing
Streamlining administrative processes 73% staff efficiency is
our top priority.
Streamlining communication 60%

Optimizing resource allocation 53%

Supporting innovation and research 53%

18
 Cities reported multiple barriers to the adoption of generative AI, citing insufficient
technical expertise (74%), lack of awareness (72%) and budget constraints (70%)

Insufficient technical expertise 74% …generative AI has

the potential to
Lack of awareness 72%
transform an entire
Budget constraints 70% industry so in order
for me to progress
Data privacy and security concerns 62% rapidly in the
upcoming years an
Lack of relevant policies or guidelines 55%
in-depth knowledge
Ethical or social concerns 47% of AI is vital
because we are
Limited access to necessary infrastructure 36% entering a period of
Lack of suitable use cases 23%
generational
change.
Note: 15 cities out of 53 reported not using generative AI 19
 Cities not using generative AI reported multiple barriers, citing lack of awareness
(93%), budget constraints (80%) and lack of relevant policies or guidelines (67%)

Lack of awareness 93%

…the biggest
Budget constraints 80% question is how can
Lack of relevant policies or guidelines 67% it add value to the
community. Another
Insufficient technical expertise 60%
is the ethical
Data privacy and security concerns 53% considerations of
AI. And also, what
Limited access to necessary infrastructure 33%
will the impact of AI
Lack of suitable use cases 27% be on the economy
and jobs.
Ethical or social concerns 20%

Note: 15 cities out of 53 reported not using generative AI 20

 Cities exploring, testing or implementing generative AI reported insufficient technical
expertise (77%), data privacy and security concerns (67%), budget constraints
(64%) and lack of awareness (64%) as key barriers

Insufficient technical expertise 77% …generative AI has

the potential to
Data privacy and security concerns 67%
transform an entire
Budget constraints 64% industry so in order
for me to progress
Lack of awareness 64% rapidly in the
upcoming years an
Ethical and social concerns 56%
in-depth knowledge
Lack of relevant policies or guidelines 51% of AI is vital
because we are
Limited access to necessary infrastructure 36% entering a period
of generational
Lack of suitable use cases 21%
change.

Note: 38 cities out of 53 reported exploring, testing or implementing generative AI 21

5 Areas of collaboration

22
Cities expressed a strong willingness to collaborate and share expertise in23tackling urban
challenges, advancing generative AI knowledge and developing best practices and policies
for responsible use of generative AI

Key areas cities indicated as opportunities for collaboration and expertise sharing in generative AI included:

COLLABORATIVE SOLUTIONS FOR URBAN

DEVELOPING USE CASES AND BEST PRACTICES
CHALLENGES

• Smart security and traffic management • Transport service optimization

• Sharing and analyzing urban datasets • Automated customer service and support
• Resident engagement and input • Crime and safety analysis

ADVANCING KNOWLEDGE AND EXPERTISE IN AI POLICY, GOVERNANCE, AND RESPONSIBLE USE

• Personalizing citizen experiences • Maintaining integrity and source of truth

• Expanding labour productivity • Ethical use and clear policies
• Emerging new business models • Transparency, security, and impacts on the future of cities

Note: Results from thematic analysis of narratives shared by cities in a survey conducted across 53 cities across the globe 23
6 Methodology

24
Methodology

OBJECTIVE DATA ANALYSIS

The primary objective of this research is to gain comprehensive insights into the Data analysis encompassed both descriptive and qualitative approaches. Quantitative data from both
perspectives, practices, and readiness of city administrations across the globe in relation to survey phases were subjected to statistical techniques, leading to the computation of descriptive
Generative Artificial Intelligence (generative AI) adoption and integration. statistics such as frequencies and %ages. These statistics were employed to summarize responses
to multiple-choice questions. Concurrently, qualitative analysis involved a sensemaking approach
applied to narrative responses obtained from the surveys to discern recurring themes, in particular to
SURVEY DESIGN two questions: “What questions do you have about the opportunities and impact of Generative AI on
cities?” from the mayors, and “What aspects of Generative AI do you want to collaborate on or get
expertise from other cities?” from city staff.
The research process was conducted in collaboration with the Centre for Public Impact in
two phases, each with a distinct focus:
• Phase 1: Mayoral survey: The first phase involved reaching out to 100 mayors across
the globe with a structured survey instrument. This survey sought to understand mayors' REPORT STRUCTURE
interest in generative AI, their questions about generative AI opportunities and usage, and
their recommendations for contact persons within their respective cities who could provide The results of this research are organized into four main sections:
further insights. The survey was designed to gather initial information about generative AI • Mayors' Interest in generative AI: This section presents findings related to mayors' levels of
awareness and interest at the mayoral level. interest, questions, and initial recommendations regarding generative AI.
• Phase 2: City-staff survey: The second phase targeted city staff members within cities • Cities' Use of generative AI: Here, we delve into the status of generative AI integration within
that responded to the mayoral survey. A comprehensive survey instrument was cities, including the presence of designated leads, sectors of application, specific use cases, and
administered to gather more detailed information about the city's generative AI activities. the presence of guidelines and policies.
This included inquiries about the presence of a generative AI integration lead, the current
• Opportunities and Barriers: This section highlights opportunities identified by city staff
status of generative AI exploration and usage, specific applications in different sectors,
members for generative AI usage, as well as the barriers and challenges they face or anticipate.
existing guidelines and policies, training initiatives, opportunities, barriers, ethical
principles guiding usage, and ongoing generative AI projects. The survey also explored • Areas of Collaboration: The final section discusses potential areas of collaboration and
potential areas of collaboration and future opportunities. knowledge-sharing among cities in the context of generative AI integration

DATA COLLECTION
LIMITATIONS AND ETHICAL CONSIDERATIONS
Both the surveys were distributed electronically and collected responses were stored
securely. The data collection period spanned mid Sep to mid Oct 2023. Contact information
The survey responses are subject to self-reporting bias, and the representativeness of the sample
for city staff members (for Phase 2 survey) was obtained from the mayoral survey
is dependent on the willingness of mayors and city staff to participate. And thus, findings are
responses. The response rates for Mayoral survey was 82% and for the City Staff survey
based on the responses of participating cities and may not be generalized to all cities worldwide.
was 65%. All data were collected and stored in compliance with ethical guidelines, ensuring
the confidentiality and privacy of survey participants while data analysis was conducted.
25