2024/4

TESEV
BRIEFS

Urban Governance
and Responsible
Artificial Intelligence
Interaction for
Local Democracy

TESEV Briefs aim to share with the public different opinions and recommendations on issues that are under TESEV’s working areas.
 TESEV BRIEFS 2024/4

Itır Akdoğan Every time a new page opens in the field of technology,
technology fans and optimists create fantasies that a
wonderful era has begun in which life will never be the
After graduating from Notre same again. The technophobes and pessimists also
Dame De Sion French lycée, she
create fantasies that life will never be the same again,
completed her BA in communication
at Galatasaray University, MA in but on the contrary that technology will bring about the
international politics at Université end of humanity. While both fantasies inherently are
Libre de Bruxelles (ULB), and PhD in frustrated1, in conceptual debates social determinism and
social sciences at Helsinki University. technological determinism, that are the ideas of society
She is an international e-democracy changes technology and technology changes society,
consultant with projects she worked clash. The alternative to this conflict is interactionism.
for in Northern Europe with the
Interactionism is the idea that if there is to be change, it
Ministry of Justice in Finland, and
in Africa with German Corporation will be through the interaction of society and technology.
for International Cooperation (GIZ). In discussions on technology and democracy within
She is the author of the book “Dijital this alternative framework, one can ask whether more
Politik Fanteziler” (Digital Political democratic processes can emerge from the interaction
Fantasies). She continues her of both fields. In this policy brief, we discuss how the
academic work with publications in interaction of responsible artificial intelligence (RAI) and
the field of e-democracy. Itır Akdoğan
urban governance might democratize cities, with a closer
is the research director of TESEV
since August 2015. look at Türkiye.

1
 TESEV BRIEFS 2024/4

As public institutions, as well as the private sector, began to use this

technology in their decision-making and implementation processes,
democratic concerns grew. The expectation of responsibility in the
use of AI for the benefit of society and for the benefit of people has
increased.

Artificial intelligence (AI), whose foundations Responsible Artificial

were laid in 1950 when British mathematician Intelligence
Alan Turing asked if machines could think and
In this context, RAI can be briefly defined as
named by John McCarthy at the Dartmouth
the ethical, democratic and legal framework
Conference in 1956, has been developing with
that is expected to be followed in the design,
its fluctuating popularity ever since. While it is
development and use of AI, with respect to
used in many sectors in the making, the masses
human rights, society and the environment.
started to incorporate ChatGPT directly into their
Prior to ChatGPT, in 2016, leading AI technology
daily lives in 2022 when Open AI launched the
companies Amazon, Google, IBM, then named
AI chatbot. The fantasies above, both good
Facebook and Microsoft came together to form
and bad, have since spread just as rapidly as
the Partnership on AI. The network which later
the use of AI. As public institutions, as well as
included media outlets, non-profit organizations
the private sector, began to use this technology
and academia, identifies 6 prioritized areas on
in their decision-making and implementation
its website to examine the opportunities and
processes, democratic concerns grew. In the case
threats AI presents for society and to create
of robots that have human skin-like textures and
recommendations and resources for RAI3:
use facial expressions similar to human emotions
(Turing had already defined intelligence as Safety-Critical AI - ensuring that
external performance, not internal, and based AI decisions in safety-critical areas
his argument on the imitation of humans by such as healthcare, transportation,
machines2), the need for a more sensitive etc. are safe, reliable and ethical for
ethical and legal framework than in the previous the people who will be affected,
technological advances arose. The expectation
Fair, Transparent, Accountable AI -
of responsibility in the use of AI for the benefit
ensuring that the data and decisions
of society and for the benefit of people has
made by AI from that data in areas
increased.

2
 TESEV BRIEFS 2024/4

such as health, education, security, AI and Social Good - AI can increase

sustainability, etc. are fair, explainable social good in areas such as education,
and accountable so that they are housing, public health, sustainability.
not prejudiced, biased or blind, Developing ideas, some of which
may be crazier and some of which
AI, Labor and the Economy -
may be more plausible, through
expanding the opportunities that
multi-stakeholder collaborations.
come with the use of AI to create new
jobs and ways of working, minimizing As this network led by private sector begins its
disruption while opening up space work, individual technology companies are also
for competition and innovation, starting to publish their own RAI declarations.
Google, for example, published its AI principles
Collaboration between People
in 2018 as follows; be socially beneficial, avoid
and AI Systems - facilitating the
creating or reinforcing unfair bias, be built and
collaboration of people and AI systems
tested for safety, be accountable to people,
to meet the needs and capabilities of
incorporate privacy design principles, uphold
people to make faster, more accurate
high standards of scientific excellence, and be
decisions, for example in areas such
made available for uses that accord with these
as medical diagnostics and traffic,
principles4. Over time, as AI and its areas of
Social and Societal Influences of AI use evolve, these principles are updated. For
- AI can affect people and society in example, Intel’s RAI list5 opens with human
areas such as privacy, democracy, rights, emphasizes the human dimension
justice, human rights, facilitating their with concepts such as inclusion and equity,
decisions in these areas, while at the and emphasizes the environment. Nokia Bell
same time deliberately manipulating Lab’s6 definition of RAI includes the principle of
them. Reflection and open dialog using AI to strengthen society and democracy.
about the possible visible and invisible Companies are also publishing transparency
impacts of AI on people and society, reports. Microsoft, for example, in its latest 2024
RAI transparency report, published information

3
 TESEV BRIEFS 2024/4

on the tools it offers to its customers for their RAI

awareness and adaptability, implement AI ethics
development, the multi-stakeholder networks throughout the technology design’s all stages,
on RAI participated in and learned from, the establish AI ethics monitoring mechanisms,
R&D and academic research on RAI supported and ensure transparency in AI processes to
around the world, the expanded RAI community, gain the trust of stakeholders and society as
and internal RAI capacity building activities7. well as accountability. ISO also underlines that
organizations that have adopted the RAI should
T he Internatio nal O rganizatio n fo r
share their principles visibly for a wide audience.
Standardization (ISO) defines RAI with these
principles: datasets used for training the AI International organizations to which governments
system must be given careful consideration to are affiliated to are also acting for the responsible
avoid discrimination; these systems should be use of AI. The United Nations (UN), for example,
designed in a way that allows users to understand established an AI Working Group in 2020.
how the algorithms work; AI systems should avoid In 2022, the UN Educational, Scientific and
harming individuals, society or the environment; Cultural Organization (UNESCO) and the Office
developers, organizations and policymakers must of Information and Communication Technology
ensure AI is developed and used responsibly; (OICT) jointly developed 10 principles of RAI
AI must protect people’s personal data, which that are adopted across the UN system. These
involves developing mechanisms for individuals to are: do no harm; defined purpose, necessity
control how their data is collected and used; and and proportionality; safety and security;
inclusivity should be ensured by including diverse fairness and non-discrimination; sustainability;
perspectives in the process8. ISO also states right to privacy, data protection and data
that while developing RAI, companies should governance; human autonomy and oversight;
foster collaboration across all disciplines such transparency and explainability; responsibility
as policy, technology, ethics and social advocacy and accountability; inclusion and participation9.
to ensure multifaceted perspectives, prioritize The AI Advisory Group established by the UN
ongoing education on AI at all levels to maintain Secretary-General in 2023 drafted the interim

4
 TESEV BRIEFS 2024/4

report, expected to be finalized at the Future website13 correspond to the above criteria. The
Summit in September 2024, that includes United States of America launched its national
the principles of inclusivity; public interest; AI strategy in 2023, the same year its political
data governance; universal, networked and declaration on responsible military use of AI
multistakeholder; and international law. and autonomy in the military as well as civilian
Moreover, the Organization for Economic use is published14. In the Global RAI Index,
Cooperation and Development (OECD) which assesses 138 countries on the criteria of
established the Global Partnership on AI (GPAI)10. government actions, non-state actors’ initiatives,
The partnership, which currently includes 29 human rights and AI, national RAI capacity and
OECD member countries (and others interested RAI governance, the most advanced country is
in joining the network), has 4 working groups the Netherlands15. The Index draws attention
focusing on RAI, namely data governance, to the fact that despite the rapid expansion of
the future of work and, innovation, and the use of AI all over the world, the majority
commercialization. The European Union (EU) of countries do not implement RAI, and that
is another policy actor actively working on the the population of countries where measures
responsible use of AI within the framework to protect and promote human rights in the
of the Declaration of Cooperation signed by context of AI are not taken is close to 6 billion.
member states and Norway in 2018. The AI Act11
which became effective as of August 1, 2024, Urban governance and RAI
and is internationally recognized as the first AI
We see that AI is rather approached with caution,
law, classifies and regulates AI according to its
where its risks are emphasized rather than its
risks. For example, unacceptable risks such as
advantages, and that RAI is basically a set of
social scoring systems or manipulative AI are
principles to manage this technology in a way
now banned in the EU.
that no one is harmed. Similar to the principles
Among the governments, China and Canada have of urban governance. In urban governance,
been pioneers in RAI field, starting their efforts too, the concepts of cooperation, inclusiveness,
in 201712. The 12 points in the RAI declaration participation, accountability and transparency
published on the Canadian e-government come to the fore for data driven decision-making

5
 TESEV BRIEFS 2024/4

Can RAI improve local democracy by facilitating inclusive data

collection, analysis and sharing plans, open data policies,
diversification of the data ecosystem, and strengthening
collaborations with different stakeholders of the city?

processes and practices to be democratic. If a to Eurocities, the European city organization,

municipality makes these principles a priority based on the exemplary model developed by
for governing the city, these priorities also apply Helsinki and Amsterdam, a leading city in the
to the use of technology in the governance of use of AI in urban management, European cities
the city. As in almost every area of the global are developing a model for data transparency
agenda, cities are developing their own policies that can be shared and copied with the principle
and practices of RAI. Cities that stand out with of interoperability. The model, developed in
their data driven urban management practices, cooperation with Barcelona, Bologna, Brussels,
such as Boston16 and Helsinki17, for example, Eindhoven, Mannheim, Rotterdam and Sofia,
publish on their websites the principles of RAI aims to prevent the misuse of data18. Can
that they declare they will comply with while collaborations developed with technology for
managing both the city and the municipality the responsible information sharing and learning
as a public institution. Both cities include the from each other democratize urban governance?
principles stated in the framework above in their
Can RAI, which includes the democratization of
RAI lists. Helsinki, for example, emphasizes focus
data governance, improve local democracy by
on people with human-centredness and human
facilitating inclusive data collection, analysis and
oversight over AI operation. Can municipalities
sharing plans, open data policies, diversification
democratize their in-house governance with RAI?
of the data ecosystem, and strengthening
While municipalities’ own governance principles collaborations with different stakeholders of
are reflected in their RAIs, we also see that RAI the city?
processes strengthen cooperation. According

6
 TESEV BRIEFS 2024/4

What is important here is that municipalities consider technology

not only as a service facilitator but also with its democratizing
dimension.

In its definition, ISO emphasizes that institutions focused more on infrastructure investment and
should make their RAI principles visible in a digitalization of services. Therefore, while Türkiye
way that is easily accessible to the masses. If a has a very high score on UN e-government
municipality transparently shares the principles reports19 with its success in digital infrastructure
of how it will manage AI in service, aid, urban and services, it is listed among the not free
planning decisions and implementation, and countries in Freedom House’s Internet Freedom
the data, technology and framework required index20.
for AI to work, can it also provide information
about other decision-making processes of that Local governments and
municipality? Can access to this information RAI in Türkiye
democratize urban governance? Türkiye is one of the countries that started
These questions can be diversified. The working on the AI framework relatively
answers, however, always start with the same early21. Some companies in the banking and
sentence: if the municipality wants to. Once telecommunications sectors (such as Türkiye İş
the interaction between urban governance Bankası, Turkcell) publish their RAI frameworks.
and technology begins, it starts the process As one of the 29 member countries of the
of evolution for both. What is important here aforementioned OECD Global AI Partnership,
is that municipalities consider technology not Türkiye’s National Artificial Intelligence Strategy22
only as a service facilitator but also with its covering the years 2021-25 also includes the RAI
democratizing dimension. Türkiye, for example, principles of the above-mentioned international
has missed the democratic dimension in organizations. These principles are listed as
previous digital transformation processes. In proportionality, safety and security, fairness,
the interaction between technology and the privacy, transparency and explainability,
public administration, the public administration responsibility and accountability, data
sovereignty and multi-stakeholder governance.

7
 TESEV BRIEFS 2024/4

The strategy document states that a Public government institutions and organizations will
AI Ecosystem will be established under the be increased to 1,000 people, at least 50,000
coordination of the Department of Big Data employees in central and local governments will
and AI Applications of the Presidential Digital be provided with awareness training, and at least
Transformation Office and AI and analytical 250 municipalities will be ensured to make use
projects of local governments will be supported. of AI technologies within the scope of smart
The democratic dimension, however, seems to city applications. We see that in the articles of
be absent in the 6 main priorities of the strategy. the strategy covering local governments, the
We observe that 1 measure and 1 objective in priorities are increasing the human resource
the document directly concern municipalities capacity of municipalities in the field of AI and
through smart cities, while 3 measures and 2 expanding its use in services. There is, however,
objectives also cover local government public no strategy for local governments to determine
institutions and organizations beside central and share the principles to be followed when
public institutions. Accordingly, the measures using this technology, there is no reference to RAI.
include dissemination of open government data When we look at the unions of municipalities,
studies, creating an inventory for field experts although they are increasingly including AI in
with AI and for advanced analytics projects and their training programs (such as The Union of
reporting regularly for the parties to exchange Municipalities of Türkiye) or periodicals (such
experience with each other, implementing as the Marmara Municipalities Union), we see
thematic programs to raise awareness of the that this is still at the level of raising awareness.
potential impacts and competence needs of The Union of Municipalities of Türkiye’s current
AI in central, dissemination of the use of AI strategic plan, for example, does not include AI.
applications and related technologies in smart
As for the AI policies of municipalities, the
city services. In the objectives, it is written that
priority is likewise on awareness raising. Some
employment in the field of AI in central and local
of them add the words innovation and artificial

8
 TESEV BRIEFS 2024/4

We may say that municipalities in Türkiye do not yet take AI as a

democratizing factor and do not yet open space for the democratic
dimension of technology in planning their resources.

intelligence to the names of their training has no role in, for example, internal governance.
centers and organize capacity building activities We may say that municipalities in Türkiye do not
especially for young people (e.g. Başakşehir, yet take AI as a democratizing factor and do not
Ümraniye, Nilüfer, Kütahya, Gaziantep yet open space for the democratic dimension of
Metropolitan). Some municipalities are opening technology in planning their resources. In other
smart city or artificial intelligence directorates words, the interaction of urban governance and
in their own institution or in their affiliates RAI in the focus of this discussion seems not yet
(e.g. İstanbul Metropolitan, İzmir Metropolitan begun.
affiliate Izmir Innovation and Technology Inc.).
How can the interaction of
We also see that municipalities increasingly urban governance and
use AI in municipal services. More and more Responsible Artificial Intelligence
municipalities are using AI for transportation be democratizing?
services (e.g. Konya Metropolitan), traffic
management (e.g. Ankara Metropolitan), For municipalities that want to run more
automation in disaster management and democratic cities, on the other hand, RAI
digital twin applications (e.g. İstanbul, Balıkesir might be a democratizing dimension of
Metropolitan), or communication and help digital transformation and an opportunity to
desk line management (e.g. Başiskele, Bağcılar). incorporate democratic governance principles
However, we do not see the RAI principles on the into urban governance. Some recommendations
websites of municipalities using AI. In the 2023 for that might include;
edition of the national Smart Cities Index, the AI policy should not be limited
top five ranked municipalities, namely Konya, to improving service quality;
İstanbul, Bursa, Kütahya and Erzurum, also fail in municipalities can do more with AI.
publishing RAI principles on their websites. When If they include democratic dimension
we ask employees of a municipality that use AI in their digital transformation
extensively, we learn that the organization’s AI policy, RAI might facilitate the
policy is more about improving service quality democratization of municipalities’
and time management of employees, but that it internal and external governance.

9
 TESEV BRIEFS 2024/4

The democratic dimension Technology itself should also be

of AI should be opened to a included in AI policy decisions;
multidisciplinary discussion; AI utilizing technology itself in
development and implementation technology policy decisions can also
processes, which involve many facilitate more accurate and creative
disciplines in themselves, can bring decisions. As in the example of city
together people and institutions collaborations above, municipalities
from many different disciplines can access more advanced AI uses
with different knowledge and at less cost through models to
experiences around the same table. be developed with the principle
Such a space for discussion alone is of interoperability facilitated by
democratizing. Municipalities can use technology. Municipality Unions may
this opportunity to discuss what AI has facilitate practices that will spread
to offer to the city for local democracy. the principle of interoperability.
Bottom-up decision making Municipalities should define and
processes should run; technical AI publish RAI principles; especially
experts of the municipality should municipalities that invest in data-
be included in the process when driven decision making processes,
municipal AI policies and practices open data and the use of AI can
are decided. In processes limited to prepare and publish their RAIs in a
high level decision makers without AI multi-stakeholder and participatory
expertise, the knowledge of what can way as soon as possible, in a
and cannot be done with AI remains way that also interacts with the
incomplete. However, AI experts have principles of urban governance.
the technical knowledge required
for more accurate decisions. This
can facilitate both better-informed
decision making and increased
participation within the organization.

10
 TESEV BRIEFS 2024/4

NOTES
1. For a comprehensive discussion on 13.https://www.canada.ca/en/
technology - social change fantasies government/system/digital-
see Akdoğan, 2014, Dijital Politik government/digital-government-
Fanteziler, İletişim. innovations/responsible-use-ai/
principles.html
2. For the AI journey see Kissinger,
Schmidt, Huttenlocher, 2021, The Age 14.https://www.state.gov/political-
of AI, John Murray Press. declaration-on-responsible-military-
use-of-artificial-intelligence-and-
3. https://partnershiponai.org is
autonomy/
where you can find more detailed
information for each item, here the 15.https://www.ictworks.org/global-
statements on the website are briefly index-on-responsible-ai/ Türkiye is not
quoted. on the list.
4. Google AI Principles – Google AI 16.https://www.boston.gov/sites/default/
files/file/2023/05/Guidelines-for-
5. https://www.intel.com/content/
Using-Generative-AI-2023.pdf
dam/www/central-libraries/us/en/
documents/2023-12/responsible-ai- 17.https://www.hel.fi/en/news/helsinki-
principles.pdf has-determined-ethical-principles-
for-the-responsible-use-of-data-and-
6. https://www.bell-labs.com/research-
artificial
innovation/ai-software-systems/
responsible-ai/#What-is-Responsible- 18.https://eurocities.eu/latest/nine-cities-
AI- set-standards-for-the-transparent-use-
of-artificial-intelligence/
7. https://query.prod.cms.rt.microsoft.
com/cms/api/am/binary/RW1l5BO 19.https://desapublications.un.org/sites/
default/files/publications/2022-09/
8. https://www.iso.org/artificial-
Web%20version%20
intelligence/responsible-ai-ethics
E-Government%202022.pdf
9. https://unsceb.org/principles-ethical-
20.https://freedomhouse.org/country/
use-artificial-intelligence-united-
turkey/freedom-world/2024
nations-system
21.For AI ecosystem development see
10.https://gpai.ai/
Turkish Artificial Intelligence Initiative,
11.https://artificialintelligenceact.eu/the- https://turkiye.ai/homepage/
act/
22.https://cbddo.gov.tr/en/nais
12.OECD.AI https://oecd.ai/en/
dashboards/overview

11
How to Cite:

Akdoğan, Itır. 2024. “Urban Governance and Responsible

Artificial Intelligence Interaction for Local Democracy” TESEV
Briefs 2024/4.
https://www.tesev.org.tr/en/research/urban-governance-
and-responsible-artificial-intelligence-interaction-for-local-
democracy/

Copyright © August 2024

All rights reserved. No part of this publication may be reproduced by electronic or
mechanical means (photocopies, downloading, archiving, etc.) without the permission of
the Turkish Economic and Social Studies Foundation (TESEV).
The views expressed in this publication are those of the authors’, and may not correspond in
part or in full to the views of TESEV as an institution.
TESEV would like to thank the Friedrich-Ebert-Stiftung Turkey Office for their support for this
publication.