Committente: Comitato Europeo delle Regioni (CdR) (all’interno del contratto quadro n. CDR/2021/C3/1/SEDEC/LOT2/1)
Periodo: 2024
URL: AI and GenAI adoption by local and regional administrations
Descrizione:
Artificial Intelligence (AI) is integrated in several sectors of society and its role and influence on economic and social domains are expected to increase in the years to come. The urge for a regulatory framework to support its assimilation became evident at European level. For this reason, the European Union adopted the AI Act (Regulation 1689/2024), which will be an influential reference in guiding the adoption and implementation of AI at European level and beyond due to its innovativeness and scale of application.
The recently adopted European regulation offers a definition of AI, a term coined in the 1950s but whose boundaries are not conclusively defined. In the AI Act, the definition of AI is not limited to a single technological application but extends to all machine-based systems capable of adaptive behaviour that share the ability to generate outputs of various typologies following the processing of received inputs. The GenAI subcategory is related to systems characterised by the ability to generate new content, supporting activities ranging from text generation to predictive analysis. In this respect, there is much potential for the fruitful exploitation of AI by the public sector. Despite to date it is noted that AI is most widely employed at national level around Europe, exploitation of AI by local and regional authorities (LRAs, also known as subnational authorities) – as the closest level to citizens – is essential to foster the spreading and public acceptance of this technology.
The benefits that AI could bring to local public administrations are varied and range from improving the efficiency of internal organisation to enhancing the interaction between the government and citizenship and to supporting decision-making processes. On the opposite side, the risks involved in the use of this technology are various and derive from operational, strategic and ethical challenges. With the aim of delving deeper into the opportunities and challenges that European subnational authorities face with respect to AI and GenAI technologies, this study draws and expands on research conducted to date on their adoption at local and regional level.
The study is built upon a systematic review of academic research, legislation and practices related to the adoption of AI at subnational level. Further analysis has been conducted through evidence-based observations stemming from primary data on the current state of play of AI adoption at local and regional level. Data was collected through an online survey targeting LRAs and launched in September 2024. The survey aimed at exploring the details of the concrete uptake of AI by local administrations, investigating aspects such as who promoted its adoption within the organisation, and their expectations and opinions on the future of AI use at subnational level. Additional primary data have been consolidated from interviews conducted to build eight case studies, as well as from experts who not only have expertise in AI but also have a deep understanding of its implementation in public administration settings. Lastly, in order to contribute to the academic and policy debate on the future adoption of AI with original, evidence-based insights, this study provides action-oriented recommendations based on the study findings and on a foresight analysis. This latter analysis is intended to offer an overview of possible developments of the AI in the near future.
The methodological approach used to conduct this study comprised the following steps:
A literature review about the state of play of AI and GenAI adoption at local and regional level, with special reference to the previous recent research results achieved in this area.
Data collection through an online survey to gather evidence on the characteristics, drivers and barriers of AI use amongst LRAs. The survey results are interpreted with a focus on the demographic size and the geographical area in which LRAs are located.
Identification and analysis of AI-based solutions that are led by local and regional administrations to explore the critical factors that contribute to the successful adoption and implementation of the technology at subnational level.
Foresight analysis based on the consolidation of (1) the AI trends identified by the LRAs that participated in the survey, (2) the Horizon Scanning for the relevant weak signals detection and (3) the Megatrend analysis.
Drafting of political and operational recommendations to the European Union, as well as the Member States and the LRAs.
The study is structured in three parts. Part 1 is divided in two main sections. The first section offers a high-level summary of research on the use of AI in the public sector. An overview of the uses of AI in the public sector and a summary of the academic research on the topic is therefore provided. Specific emphasis is posed on topical aspects such as the factors that enable or hinder AI adoption, citizen acceptance and the perceived risks of public sector AI. The second section has the objective of offering an overview of the current state-of-play of AI adoption by the public sector at the subnational level across the EU Member States and draws conclusions based on the analysis of the responses of the online survey submitted to LRAs in all 27 EU Member States. The survey results were divided for analysis into four areas to investigate AI adoption and implementation processes in relation to aspects of the internal organisation of local public authorities and their relationship with the ecosystem in which they are integrated.
Part 2 explores the factors for the successful AI adoption at the local and regional level. The section is based on the elaboration of primary data collected through qualitative research. In more detail, information is gathered from interviews with officials at the local and regional level and experts from academia and research centres. Interviewees contributed to enriching the analysis by sharing the theoretical knowledge and practical insights based on their professional experience into AI’s role in local and regional governance. The factors that contribute to the successful adoption and implementation of AI technologies are analysed through the lens of the Technology-Organisation-Environment (TOE) framework. Elements considered to be of relevance for the successful use of AI include: the volume of investments, both from the public sector (at regional, national or European levels) and the private sector; the degree of collaboration between public administration and private companies on the ground; and the concrete support provided by policy makers to digitisation projects. In addition, the importance of digital skills and a deep understanding of technological solutions by the personnel of the administrations is emphasised. Crucial also appears to be the involvement of citizens in the implementation of digital solutions. In conclusion, the potential scalability as a criterion for evaluating the success of a project is discussed.
Part 3 of the study is composed of two sections and is aimed at providing policy recommendations addressing relevant aspects of AI implementation. The first section is dedicated to the foresight analysis, the in-depth explanation of the components that comprised it and the exposition of the related results. The foresight analysis focuses on possible future trends and unforeseen developments in AI technology. The analysis benefitted from the responses received to the online survey, that were clustered in ‘trends’ and used to derive insights and implication of AI for public administrations. Together with trends, the analysis made use of ‘weak signals’ and ‘megatrends’. ‘Weak signals’ are considered as emerging signals of development with a likely future impact, analysed through the technique of Horizon Scanning. In particular, the weak signals relevant to the present research, as described in the 2021 JRC report, are: Ethical AI, Green AI, Smart cities and Edge AI. Finally, the analysis considered the ‘megatrends’, intended as long-term driving forces that are observable now and will continue to have a global impact in the future. Furthermore, it is postulated that megatrends can produce an effect on the trends, accelerating them and enhancing their impact.
Based on the development of the foresight analysis, in the second section proposals for political actions are advanced, directed to the EU, national governments of EU Member States and LRAs. The objective of the recommendations is to support a coordinated action on AI adoption for the public sector at local and regional level in the EU. Eight recommendations were developed on the basis of the integration of the results of the analyses carried out in the study and, in order to avoid redundancies, an overview of the recommendations issued so far on the implementation of AI in the public sector is provided beforehand. The recommendations cover various thematic areas, from monitoring and evaluation to regulatory support and the relationship between the political and technical layers.