


Project: Name “Building Administrative Capacities of the Western Balkans and the Republic of Moldova”
Publisher:
Network of Associations of Local Authorities of South-East Europe (NALAS)
Mr. Kelmend Zajazi, Executive Director
Co-publisher.:
KDZ – Centre for Public Administration Research
Mr. Thomas Prorok, Managing Director
Authors:
Mr. Christian Rupp, International Digitalization and Smart City Expert
Ms. Jana Belcheva Andreevska, Decentralization and Smart City Expert, NALAS
Authors/Contributors:
Ms. Dejana Radović, City of Podgorica
Ms. Nermina Suljević, City of Sarajevo
Mr. Branislav Misović, ALVRS
Mr. Nikola Todorovski, ZELS
Ms. Karina Donevska, ZELS
Mr. Palii Alexandru, CALM
Mr. Bosko Nenadović, SKGO
This publication is a collaborative effort of NALAS Digitalization Working Group
Design by: Brigada dizajn, Skopje

Supported by:



Disclaimer
The views expressed in this Survey, represent the opinions of the contributors, authors and editors. They do not necessarily represent the views of the ADA, KDZ or BACID.
Introduction

The digital transformation at the local level is a vital process for fostering economic growth and accelerating the European Union integration of countries in the Western Balkans and Moldova. This factsheet builds on the study Empowering Progress with Digital Transition in Western Balkan and Moldova Local Governments.
The factsheet “Artificial Intelligence in Local Government: Driving Innovation, Bridging Gaps, and Shaping the Digital Transition in the Western Balkans and Moldova”focuses on this key component for driving progress and accelerate the transformation process - the topic of artificial intelligence (AI) in local government. Along the benefits of AI that are widely discussed, the associated risks, which call for regulations governing its use are expressed.
AI has already infiltrated nearly every aspect of life, bringing profound changes that will shape the future of local innovation and influence the fairness of the digital transition. As AI continues to evolve rapidly, its impact on industries, economies, and communities grows stronger, presenting both opportunities and challenges.
To elaborate on this factsheet, the authors employed desktop research and actively engaged members of the NALAS Digitalization Working Group, who provided valuable insights on all topics covered. The study also incorporates measurements and analyses conducted by the EU for the Western Balkans and Moldova, alongside best practices that have demonstrated successful outcomes and are applicable or recommended to local governments within the EU.
In public administration, AI can revolutionize operations by streamlining internal processes, improving decisionmaking, and enhancing the delivery of services to citizens. Local governments can leverage AI to offer more efficient, transparent, and citizen-centric services, aligning with modern expectations.
To fully embrace the potential of AI, there are significant opportunities to invest in talent development and to share knowledge and experience with peer local governments, both within the region and across the EU. Such collaboration can strengthen local capacities, build networks, and ensure that the benefits of AI are accessible and equitable.
Local governments that lack access to skilled AI talent or robust infrastructure risk falling behind in this transformative era. The gap between AI-ready and those struggling to keep pace could widen, intensifying existing inequalities in digital development. However, AI also holds the potential to act as a powerful driver of change. By focusing on place-based solutions, local governments can harness AI to address unique community needs, promoting innovation and accelerating the digital transition.
AI applications, such as chatbots, are already being used, but the potential of AI extends much further and depends on the availability of local data sets and their analytics, which must be directed towards delivering higher-quality solutions for citizens, tailored to their needs at the local level.
The authors believe that this factsheet will contribute to the next steps in digital and green transformation at the local level, helping to harness best practices, recommendations, and knowledge to bridge the gap in the quality of life for citizens of the Western Balkans and Moldova through advanced technological solutions.
Lastly, the authors extend their sincere gratitude to the teams at KDZ and NALAS for the opportunity to produce this factsheet and present it to the public.

ARTIFICIAL INTELLIGENCE AND DATA ECONOMY FOR MUNICIPALITIES

The rapid and diverse development of artificial intelligence (especially after the market launch and free provision of ChatGPT in November 2022) has created a broad scope for potential applications in public administration. From the automation of processes and recommendations for decision support to real-time decision-making and the improvement of administrative services there are numerous areas of application to choose from. But which fields of benefit and application are identified for the use of artificial intelligence in public administration and how can these be utilised both effectively and efficiently?
In some areas of public administration, the use of AI is already part of everyday life.
For example, automated application and invoice processing, review of documents and preparation of judgements in mass proceedings (e.g. OLGA1 at the Augsburg Higher Regional Court and FRAUKE2 at the Frankfurt Regional Court), meeting support (e.g. F133 in Baden-Würtemberg), chatbots to relieve the burden on citizen services (e.g. WienBOT4 or LUMI5 in Heidelberg), predictive policing for better deployment planning by security authorities, optimisation of traffic management, relief in personnel management, simplification of resource planning, fraud detection, better risk management and automated translations are used as a matter of course.
Language models can be used to visualise the effects of laws in enforcement and the associated need for action in different areas of an authority. The adoption of new laws has a wide range of effects on the administration. The authorities responsible for implementing benefits are particularly affected. This results in time-consuming and error-prone interpretations, manual adjustments to follow-up documents such as implementation instructions and high training costs.
AI to solve challenges in local authorities
The German AI Institute for Local Authorities - URBAN.AI6 is working on using AI to solve a wide range of challenges in local authorities. With the help of local universities and AI research institutions, municipalities can develop concrete use cases to become the smart cities of tomorrow. With the expertise of the municipalities and the technical implementation skills of the partners, prototypes can be quickly created and tested to make the benefits of AI immediately tangible in the municipalities.
1 https://de.newsroom.ibm.com/2022-12-07-OLG-Stuttgart-pilotiert-IBM-Massenverfahrensassistenten-zur-Fallbearbeitung-inDieselabgasverfahren
2 https://justizministerium.hessen.de/presse/pressearchiv/ki-projekt-codefy-am-landgericht-frankfurt-gestartet
3 https://www.baden-wuerttemberg.de/de/service/alle-meldungen/meldung/pid/kuenstliche-intelligenz-in-der-verwaltung
4 https://www.wien.gv.at/bot/
5 https://www.heidelberg.de/Digitale-Stadt/startseite/projekte/ki-buergerassistenz+lumi.html
6 https://urban-ki.de/

AI
for urban planning & (geo)databased infrastructures
AI technologies automate the updating of cadastral data and the creation of condition maps and support the recognition of vegetation and sealing types for climate and environmental analyses.
AI for mobility planning & control
By integrating remote sensing data and crowd data, AI-based systems optimise traffic control, reduce congestion and CO2 emissions and improve the user experience in traffic.
AI for environmental planning, climate protection & climate impact adaptation
AI-supported evaluations of remote sensing data enable the analysis of municipal properties and the adaptation of infrastructures to climate change by recognising sealed areas and types of vegetation.
AI for buildings, supply & disposal
Automated processes using AI improve the updating of cadastral data, the creation of condition maps and the analysis of building structures, thereby promoting energy efficiency and sustainability.
AI for civil protection & civil security
AI-supported systems and robotic technologies support emergency services, improve data analysis and decision-making in rescue operations and promote knowledge transfer through prototypes and demonstration platforms.
AI for administrative processes & citizens
AI-based applications increase the efficiency and quality of public services, reduce the workload of employees and enable barrier-free, user-friendly interactions with citizens through conversational interfaces and voice assistants.
Local public transport is also using AI in various areas to improve its offers and services, increase internal efficiency and improve safety (e.g. KARL7). The use of AI for predictive maintenance of the vehicle fleet, timetable optimisation, customer interaction, video surveillance at train and bus stations, passenger forecasts, staff deployment planning and automation in rail and bus transport are just a few examples (e.g. DAKIMO8).
Large language models also help to generate helpful text outlines and document drafts based on text bodies and file locations, which are not yet fully developed standard documents. Text summaries and translations are further valueadded services of these language models (e.g. DEEPL9).
7 https://kompetenzzentrum-karl.de/ki-im-einsatz/leitstelle/
8 https://dakimo.server.de/
9 https://www.deepl.com/

Importance of
change
management,
the
legal framework and ethical principles
AI systems can perform certain activities more efficiently. Therefore, the introduction of AI in administration will inevitably bring about changes for employees in the public sector and will lead to a reorientation in the fields of activity.
It is not only possible fears of job losses and changes that play a role here, but also opportunities that can arise from expanding skills and focusing on more demanding tasks as well as the challenges of demographic change. The importance of change management processes is also emphasised in order to ensure the smooth integration of AI into existing workflows and to promote employee acceptance.
At the centre is always the question of attention how to ensure that AI applications comply with ethical principles, the legal framework of public administration and the free and democratic basic order. And therefore, how does the increasing use of AI solutions affect personnel in public administrations and what changes can be expected both through and in the change management processes? What competences are therefore required in the offices and authorities in order to successfully manage the use of AI and ward off any dangers without being exposed to lock-in effects or dependencies?
„Successful digital transformation does not come from implementing new technologies but from transforming an organisation to take advantage of the possibilities that new technologies provide. Besides leading the change, this also requires that all people in an organisation-leadership, IT professionals, employees in other divisions obtain the skills to embrace technology.“ 8
One of the first and therefore oldest ‘AI learning modules’ is a portal of the Finnish EU Presidency from 2018 called “ElementsOfAI” 11 which is still regularly updated and available in around 30 languages, except from the Western Balkans. NALAS could be easily bring the Elements of AI courses to all their partners in their native language.
Linking the use of technical tools with social structures and mechanisms necessitates a sharp understanding of what tools can and cannot achieve. It requires a clear definition of the interface between technology and social structures, including resolving questions of agency and control. It also entails a sufficiently defined focus for technical tools to deliver what is expected from them. This means tools must be well understood and evaluated as fit for purpose. And we must equally expect a clarity of design and process from the social structures that make up the system of guardrails, as well as a functioning interface. More challenging but also potentially more useful would be socio-technical setups that are capable of iteration, adaption, and learning - so that the system of guardrails not only helps individual decisionmakers to improve but also the system itself to evolve and progress as experience accrues and contexts become clearer.
Digital leadership in public administration
The digital transformation is a design task that does not focus on technology, but requires a holistic view of organisation, people and technology. In order to successfully exploit the potential of the digital transformation of public administration, the skills of employees are increasingly taking centre stage. Skills are seen as a relevant driver of digitalisation in the public sector. Local authorities need to develop as well as knowledge about the potential and risks of big data. Municipalities’ AI and smart city activities require the establishment and continuous (further) development of expertise in urban data. However, in view of existing mandatory tasks and specialised procedures, smaller and medium-sized municipalities in particular are faced with the question of how such an additional challenge can be overcome and how integration into the administration can be ensured.12
10 European Commission (2016) eGovernment Benchmark 2016: A turning point for eGovernment development in Europe (p 76f) http://publications. europa.eu/resource/cellar/b3627b36-b212-11e6-871e-01aa75ed71a1.0001.01/DOC_1
11 https://www.elementsofai.com/
12 Rupp, C.: Wie der öffentliche Sektor KI schon einsetzt (S 29-31), Fachzeitschrift Innovative Verwaltung, Ausgabe 11/2023, Springer Fachmedien Wiesbaden (ISSN 1618-9876), https://www.springerprofessional.de/wie-der-oeffentliche-sektor-ki-schon-einsetzt/26299462

Digital leadership in public administration is understood as “a new form of leadership that attempts to harmonise employees, technologies, the framework conditions of public administration and the wishes of citizens and companies in times of digital change”.13 Digital leaders are needed in public administration who act as drivers of innovation and create a climate that enables the development of a digital mindset. They are leaders to establish the local digital ecosystem and promote partnership for development.
Skills, education, campuses and training platform
In terms of transfer to the day-to-day work of public administration, the identified competences can be built up through educational events such as workshops. The “Digital Leadership” centre in Germany14, for example, offers such events in which agile project management methods such as Scrum and problem-solving approaches such as design thinking are taught. However, it is emphasised that managers should not simply adopt what they have learned, but must apply it to their own situation in a self-reflective manner.
There is also informal learning, which promotes independent exploration of new technologies. In addition, employees can be trained as digital coach who act as multipliers and teach their colleagues. Communities of practice (CoP) and homogeneous focus groups offer the opportunity for managers to discuss specific use cases together and develop solutions to problems.
In the area of e-learning, the development of an education and training platform in the form of a Massive Open Online Course (MOOC) solution is recommended. Such a platform enables various universities to offer educational content that can be supplemented by blended learning (face-to-face and e-learning) and simulation games.
A good example is the AI-Campus 15 in Germany. A learning platform for artificial intelligence with free online courses, videos and podcasts in various topics of AI and data literacy. All learning opportunities are available free of charge. In addition, AI Campus Originals are openly licensed (CC BY-SA 4.0) 16 . A cooperation between NALAS and the AI-Campus is recommended in order to offer these courses to Western Balkan and Moldova.
Fellowship programmes are also recommended to ensure the transfer of newly learned methods and working methods into everyday working life. Experts from the private sector can teach employees new ways of working as part of projects within the authority.
At the fellowship Tech4Germany 17, for example, the best talents and experts from the fields of product, design and engineering work with digital pioneers from federal ministries and authorities on specific digital projects. In interdisciplinary teams, they develop a deep understanding of problems and users using agile and user-centred methods and create prototype software products within three months.
13 Kusanke K., Kendziorra J., Pilgenröder S., Winkler T.J.: Maßnahmen für den Aufbau von Digital Leadership - Kompetenzen für Führungskräfte im öffentlichen Sektor, HMD Praxis der Wirtschaftsinformatik (2024) 61:202–219, https://doi.org/10.1365/s40702-023-01026-3
14 https://www.diefuehrungsakademie.de/kompetenzzentrum-digital-leadership
15 https://ki-campus.org/front?locale=en
16 https://creativecommons.org/licenses/by-sa/4.0/
17 https://digitalservice.bund.de/fellowships/tech4germany

Ethical use, guidelines and legislation
The challenge for the use of AI in practice is to organise it in such a way that administrative processes and decisions can benefit from it without neglecting concerns, particularly with regard to data protection and ethics, and without undermining fundamental democratic values. The effects of such changes on technology, activities, processes, organisations and ultimately the entire public sector can be guessed at, but their concrete form remains shrouded in the fog of uncertainty and the exaggerated expectations of social hype.
Guidelines are essential and NALAS could develop such guidelines as a blueprint for members. Good examples from Austria include: AI compass City of Vienna18 and the AI strategy City of Linz 19, or the Guidelines from BMAS Germany20 or BMÖLS Austria 21. These guidelines also offer a practical decision tree, criteria and measures to help you decide whether and under what conditions data-driven AI technology can be used in public administration. The checklist for ethical AI 22 in public administration in Austria also provides a brief initial overview.
For starters, successfully regulating AI requires a robust and effective administrative state, the ability for individuals to bring causes of action against culpable companies, an approach to free expression focused on people and not profit, and the political will and ability to enforce the rules we’ve already got 23. Transparency, consent, self-regulation, and limiting uses of AI: lawmakers should embrace a strategy of duties, design rules, defaults, and data collection dead ends for data processing and deployments of AI. This layered approach will more squarely address data extraction, normalization, and self-dealing and better ensure that research and development into AI advances the public good.
Table 2: Global Index on Responsible AI 24
COUNTRY INDEX SCORE
PILLAR SCORE
18 https://digitales.wien.gv.at/ki-kompass-fuer-bedienstete-der-stadt-wien/
19 https://www.linz.at/images/files/ki-strategie_stadt_linz.pdf
20 https://www.bmas.de/SharedDocs/Downloads/DE/Publikationen/a862-01-leitlinien-ki-einsatz-behoerdliche-praxis-arbeits-sozialverwaltung.html
21 https://oeffentlicherdienst.gv.at/wp-content/uploads/2023/11/Leitfaden-Digitale-Verwaltung-Ethik.pdf
22 https://oeffentlicherdienst.gv.at/wp-content/uploads/2023/11/231114_Formulare_Checkliste_ethische_KI.pdf
23 https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2024/02/ai-other-companies-quietly-changing-your-terms-service-could-be-unfairor-deceptive
24 Measuring responsible AI in 138 countries https://global-index.ai/Region-Europe

Excursus: EU AI Act (AIA)
With the AI Act 25, the EU is one of the world’s pioneers in the regulation of artificial intelligence. The AI Act ensures the safe use of AI systems, creates investment security through a clear legal framework and thus strengthens the research, development and use of AI systems. The AI Act provides for many roles (actors in accordance with Art. 3 (8)) in the AI value chain:
z Provider
z Product Manufacturer
z Authorised Representative
z Importer
z Distributor
z Deployer
There are also users and ‘data subjects’. However, these are not referred to as actors within the meaning of the AI Act. The various players take on different activities in the AI value chain. A distinction is made between ‘placing on the market’, ‘provision’ and ‘commissioning’, which are also legally defined in the AI Act.
The AI Act follows a risk-based approach in order to introduce a proportionate, effective and binding set of rules for AI systems. AI systems are categorised according to their risk potential as unacceptable, high, low and minimal risk. The AI Act defines risk as ‘the combination of the likelihood of harm occurring and the severity of that harm’ (Art. 3) and assesses in particular potential harm to individual and/or public interests (health, safety, fundamental rights, including democracy, the rule of law and environmental protection). Damage can be of a material or immaterial nature. It covers physical, psychological, social or economic damage.
General purpose AI models occupy a special position in this categorisation.
There is a clear distinction between the terms ‘AI systems’ and ‘AI models’. Not all AI models fall within the scope of the AI Act, but only GPAI models. Although GPAI models can be part of an AI system, they do not form an AI system in isolation. In order for a GPAI model to become an AI system, additional components - such as a user interface - must be added. In this case, it is referred to as a general purpose AI system (or GPAI system) within the meaning of Art. 3 (66) AIA.
The timeframe of the EU AIA - Entry into force of the AI Act: 1st August 2024 (20th day after publication in the Official Journal of the European Union).
Six months after the AI Act comes into force (1st February 2025), the practices classified as ‘prohibited’ may no longer be used. The regulations on AI competence (‘literacy’) also apply. This means that providers and operators of AI systems are obliged to take measures to ensure that their staff and other persons involved in the operation and use of AI systems on their behalf have sufficient knowledge. The provisions on general purpose AI systems must be applied twelve months after the AI Act comes into force (1st August 2025). Furthermore, the regulations on notification bodies also apply. EU-Member states are obliged to designate at least one notifying authority that is responsible for setting up and implementing the necessary procedures for the assessment, designation and notification of conformity assessment bodies and for their monitoring. The governance provisions must also be applied, whereby the EU-Commission at Union level and the EU-Member-States at national level are required to establish or designate the intended authorities and institutions. In addition, the penalty provisions apply from this date on.
24 months after the AI Act comes into force (1st August 2026), all obligations apply in principle. This means that the requirements for high-risk AI systems in accordance with Annex III (but not Annex I) and AI systems with low and minimal risk must be complied with. 36 months after the AI Act comes into force (1st August 2027), the requirements for high-risk AI systems in accordance with Annex I apply.
25 https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=OJ:L_202401689
