Conference PaperPDF Available

The use of AI in public services: results from a preliminary mapping across the EU

Authors:

Abstract and Figures

Artificial Intelligence is a new set of technologies which has grasped the attention of many in society due to its potential. These technologies could also provide great benefits to public administrations when adopted. This paper acts as a first landscaping analysis to indicate, classify and understand current AI-implementations in public services. By conducting a desk research based on available documents describing AI projects, 85 AI applications in the public sector in selected European countries have been identified and reviewed. The preliminary analysis suggests that most AI initiatives are started with efficiency goals in mind, and they occur mainly in the general public service policy area. Findings of this preliminary landscape analysis set the basis for further more in depth research and recommendations for policy.
Content may be subject to copyright.
A preview of the PDF is not available
... Many of these new applications, systems or applications are considered AI as they can conduct intelligent tasks [9]. Existing studies already highlight that there are many, varying forms of technologies and applications considered to be AI, which may not be alike [12][13][14]. What is the most fundamental for research on AI in the government, however, is that civil servants themselves may use different terms and concepts to understand and describe Artificial Intelligence. ...
... Ability of machines to carry out tasks which require human capabilities, by displaying human-like behaviour, to behave rationally, the ability to solve hard problems [22,35,46,47] AI as applications A special form of IT systems, applications or software that are capable of performing tasks that normally need human intelligence [12,13,30,38] AI as a science The general study and science behind the pursuit of making machines or computers intelligent [5,39,48] An overview of these different understandings, which at times can overlap, can be found in Table 1 below. ...
... These characteristics not only call for a redefinition of the division of labour between humans and machines (Choudhary et al. 2021) but also, and more broadly, for a complete rethinking of organizational processes, culture and relations affected by AI -especially in public settings (Mergel, Edelmann, and Haug 2019). Misuraca, van Noordt, and Boukli (2020) identified different applications in European public organizations, showing that AI is mainly adopted in public service delivery. However, and surprisingly, scholars in public management rarely investigated the topic (Kankanhalli, Charalabidis, and Mellouli 2019;Sousa et al. 2019), especially through empirical research (Dwivedi et al. 2021). ...
... Chatbots are one of the most researched ( van Noordt, and Misuraca 2019) and widespread AI applications among private and public organizations (Misuraca, van Noordt, and Boukli 2020). Chatbots' expectancies are to help public institutions in their daily relations with citizens and firms on reducing the administrative burden and increasing the quality of the communication channels (Androutsopoulou et al. 2019). ...
Article
We investigate how AI introduction affects public entities at the micro-level, hence the roles, competences and tasks of the agents involved. In doing so, we rely on the organizational design theory and we focus on a specific AI solution (chatbot) implemented within a defined microstructure, the customer service department. Using data collected through six exploratory case studies, we show how the creation of an AI team becomes a novel form of organizing that solves the universal problems of organizing. Results confirm that AI implementation is a complex organizational challenge and suggest that artificial agents act similarly to human ones.
... Ideas can be found in the areas of healthcare, transportation, education, and security (de Sousa et al., 2019), and there are many examples of how AI can be implemented to support public services (Rosemann et al., 2020). There are even projects in which AI has been integrated into public administration (Misuraca et al., 2020) and into the health sector (Yang, Ng, Kankanhalli, & Yip, 2012). Nevertheless, no findings on their acceptance have been reported. ...
... Implementation of AI in public services has the potential to increase service efficiency and service quality for citizens (Galloway & Swiatek, 2018;Kowalkiewicz & Dootson, 2019;Misuraca et al., 2020;Rosemann et al., 2020). The difference between AI-based and non-AI-based software is that AI is self-learning and thus can manage new situations without further programming (Leyer & Schneider, 2021). ...
Article
Full-text available
Interest in implementing artificial intelligence (AI)–based software in the public sector is growing. First implementations and research in individual public services have already been carried out; however, a better understanding of citizens' acceptance of this technology is missing in the public sector, as insights from the private sector cannot be transferred directly. For this purpose, we conduct policy-capturing experiments to analyze AI's acceptance in six representative scenarios. Based on behavioral reasoning theory, we gather evidence from 329 participants. The results show that AI solutions in general public services are preferred over those provided by humans, but specific services are still a human domain. Further analyses show that the major drivers toward acceptance are the reasons against AI. The results contribute to understanding of when and why AI is accepted in public services. Public administration can use the results to identify AI-based software to invest in and communicate their usage to perceive such investments' high acceptance rates.
... Robots, virtual assistants, and other kinds of agents imbued with artificial intelligence (AI) have been, and will continue to be, increasingly implemented across many industries, such as healthcare, military, and transportation, to name a few of the many areas that are being transformed by advances in these technologies (Misuraca et al., 2020). AI has also been useful in numerous different applications, from healthcare, such as evaluating and determining treatment in orthopedic surgeries (Haleem et al., 2020), to business, where machine learning drives marketing decisions (Cuzzolin et al., 2020;Rai, 2020). ...
... Thus, the reasons for which they exhibit certain behaviors are often not easily interpretable (Calder et al., 2018;Rai, 2020). Humans are not capable of replicating the processes machine learning models use to discover patterns in data (Misuraca et al., 2020), nor how these models rationalize their determinations about a system (Bennett and Maruyama, 2021). This has led to the development of new areas of research, such as "explainable artificial intelligence" (XAI; Gunning et al., 2019). ...
Article
Full-text available
In this paper, we discuss the development of artificial theory of mind as foundational to an agent's ability to collaborate with human team members. Agents imbued with artificial social intelligence will require various capabilities to gather the social data needed to inform an artificial theory of mind of their human counterparts. We draw from social signals theorizing and discuss a framework to guide consideration of core features of artificial social intelligence. We discuss how human social intelligence, and the development of theory of mind, can contribute to the development of artificial social intelligence by forming a foundation on which to help agents model, interpret and predict the behaviors and mental states of humans to support human-agent interaction. Artificial social intelligence will need the processing capabilities to perceive, interpret, and generate combinations of social cues to operate within a human-agent team. Artificial Theory of Mind affords a structure by which a socially intelligent agent could be imbued with the ability to model their human counterparts and engage in effective human-agent interaction. Further, modeling Artificial Theory of Mind can be used by an ASI to support transparent communication with humans, improving trust in agents, so that they may better predict future system behavior based on their understanding of and support trust in artificial socially intelligent agents.
... 23 The interest and uptake of these advanced forms of AI in the private and public sectors is growing. 24 Due to their self-learning abilities, the detailed decisionmaking process of machine learning algorithms is often opaque. Even the programmers do not always fully understand how a particular decision was reached. ...
Research
Full-text available
Report summarizes how Artificial Intelligence can be a new tool for corrupt practices and what can be done to meet that emerging risk.
... Literature on e-government adoption using such models propagates bias towards managerial and economic outcomes driven by NPM tenants (Cordella & Bonina, 2012) and continues to be driving AI implementations. Misuraca et al. (2020) review of 85 AI implementations in the European public sector shows 70% were driven by performance and efficiency goals, with only 30% being focussed on making the government open and none on public values. As well, the expected benefits in 56.5% are internally motivated towards organisational performance and only 27.1% towards social values (Ibid.). ...
Chapter
The use of artificial intelligence (AI) by governments represents a radical transformation of governance, which has the potential for a lean government to provide personalised services that are efficient and cost-effective. This represents the next frontier of digital-era governance (DEG), which is an extension of the traditional bureaucratic model representing digital manifestations of instrumental rationality. However, the use of AI also introduces new risks and ethical challenges (such as biased data, fairness, transparency, the surveillance state, and citizen behavioural control) that need to be addressed by governments. This chapter critiques DEG enabled by AI. The authors argue for adopting a public values perspective for managing AI ethical dilemmas. Through a cross-case analysis of 30 government AI implementations, four primary AI use cases are outlined. Furthermore, a conceptual model is developed that identifies relationships between AI ethical principles and public values as drivers of AI adoption by citizens. Finally, six propositions are outlined for future research.
... In a review of previous literature, de Sousa et al. [9] had found only two instances of NLP use in government. However in a more recent survey on governmental use of AI in the EU, Misuraca et al. [20] identified 85 implementations of such use and found out of all AI technologies, NLP was the most prevalent. ...
Chapter
This research presents a comprehensive understanding of AI in the public sector based on a review of 78 studies. The literature review indicates that an AI-analytical model and AI-based automation system are mostly used at the organizational level whilst AI-recommender and chatbot applications are implemented within the citizens’ services context. The results reveal that AI benefits such as cost reduction and decision-making improvements are accrued by governments. Further, the benefits of personalization and positive user experiences are directly useful to citizens. The review highlights that developing and adopting AI, presents two categories of challenges: AI obstacles at the organizational level, such as employees’ resistance, lack of managerial, and financial support, and second - AI dilemmas linked to citizens such as AI ambiguity, bias, and privacy. Accordingly, this study provides recommendations for further research on AI within the government and the public sector.
Chapter
The purpose of this chapter is to identify and analyze key tensions in the governance, the management, and the organization of RPA-practices in public services. Key tensions will be analyzed in an agency–structure framework and using empirical data on introduction of RPA-tools in income support services in three Swedish municipalities. The analysis will be guided by two research questions: what tensions underly RPA-integration by actors in their practices; and how these tensions can be understood based on an agency–structure framework. The findings discuss issues such as coordination dilemmas in decentralized structures, different interpretations of the legal frames, standardization versus individualization logics of the profession, trust versus control mechanisms, but also elements of municipal innovation and accommodation of case processing when RPA is integrated into income support services.
Chapter
Full-text available
The aim of this paper is to provide an overview on how artificial intelligence is shaping the digital era, in policy making and governmental terms. In doing so, it discloses new opportunities and discusses its implications to be considered by policy-makers. The research uses a systematic literature review, which includes more than one technique of data analysis in order to generate comprehensiveness and rich knowledge, we use: a bibliometric analysis and a content analysis. While artificial intelligence is identified as an extension of digital transformation, the results suggest the need to deepen scientific research in the fields of public administration, governmental law and business economics, areas where digital transformation still stands out from artificial intelligence. Although bringing together public and private sectors, to collaborate in the public service delivery, presents major advantages to policy makers, evidence has also shown the existence of negative effects of such collaboration.
Article
Full-text available
Smart information and communication technologies (ICTs) are finding their ways into public administration, and numerous smart government efforts are marking the start of a new digitalization wave in the public sector. Despite being in the early stages of development, these initiatives promise a new model for the provision of public services: smart government. Because past technical innovations in the public sector did not reach their full potential, it is crucial to know the difficulties if one is to successfully address them. We explore the perceived barriers to the adoption of smart government in an early phase of implementation. We analyzed barriers, utilizing 32 interviews with actors involved in smart government initiatives. Cluster analysis helped us to identify six barrier groups: a lack of legitimacy, a lack of legal foundations, a lack of policy coherence, a lack of technical infrastructure, cost-benefit considerations, and a lack of innovation capacity. We distinguish between organizational and institutional barriers, and discuss restrictions and implications for praxis and future research.
Article
Full-text available
This study organizes existing research on the public value of e-government in order to investigate the current state and what value e-government is supposed to yield. The two questions that guided the research were: (1) What is the current state of research on the public value of e-government? And (2) What value is e-government supposed to yield? Six, sometimes overlapping, values were found: Improved public services; improved administrative efficiency; Open Government (OG) capabilities; improved ethical behaviour and professionalism; improved trust and confidence in government; and improved social value and well-being. These six public value dimensions were thereafter generalized into three overarching, and also overlapping, public value dimensions of Improved Public Services, Improved Administration, and Improved Social Value. The improved public services dimension influences other dimensions. Hence, this literature study theorizes a descriptive and multidimensional framework that can improve our understanding of the public value of e-government from different viewpoints, and the overlap between them in actual e-government designs and implementations. Regarding the current state of research on the public value this study found a lack of research on the public value of e-government, especially, in the context of developing countries – and more importantly – a total absence of this kind of research in the Least Developed Countries (LDCs). There is also a lack of comparative studies at national, regional, and project level; and a lack of research on the generative perspective.
Article
Full-text available
Driven by the technological capabilities that ICTs offer, data enable new ways to generate value for both society and the parties that own or offer the data. This article looks at the idea of data collaboratives as a form of cross-sector partnership to exchange and integrate data and data use to generate public value. The concept thereby bridges data-driven value creation and collaboration, both current themes in the field. To understand how data collaboratives can add value in a public governance context, we exploratively studied the qualitative longitudinal case of an infomobility platform. We investigated the ability of a data collaborative to produce results while facing significant challenges and tensions between the goals of parties, each having the conflicting objectives of simultaneously retaining control whilst allowing for generativity. Taken together, the literature and case study findings help us to understand the emergence and viability of data collaboratives. Although limited by this study's explorative nature, we find that conditions such as prior history of collaboration and supportive rules of the game are key to the emergence of collaboration. Positive feedback between trust and the collaboration process can institutionalise the collaborative, which helps it survive if conditions change for the worse.
Article
Full-text available
The nascent adoption of Artificial Intelligence (AI) in the public sector is being assessed in contradictory ways. But while there is increasing speculation about both its dangers and its benefits, there is very little empirical research to substantiate them. This study aims at mapping the challenges in the adoption of AI in the public sector as perceived by key stakeholders. Drawing on the theoretical lens of framing, we analyse a case of adoption of the AI system IBM Watson in public healthcare in China, to map how three groups of stakeholders (government policy-makers, hospital managers/doctors, and IT firm managers) perceive the challenges of AI adoption in the public sector. Findings show that different stakeholders have diverse, and sometimes contradictory, framings of the challenges. We contribute to research by providing an empirical basis to claims of AI challenges in the public sector, and to practice by providing four sets of guidelines for the governance of AI adoption in the public sector.
Article
Full-text available
One of the most noticeable trends in recent years has been the increasing reliance of public decision-making processes (bureaucratic, legislative and legal) on algorithms, i.e. computer-programmed step-by-step instructions for taking a given set of inputs and producing an output. The question raised by this article is whether the rise of such algorithmic governance creates problems for the moral or political legitimacy of our public decision-making processes. Ignoring common concerns with data protection and privacy, it is argued that algorithmic governance does pose a significant threat to the legitimacy of such processes. Modelling my argument on Estlund’s threat of epistocracy, I call this the ‘threat of algocracy’. The article clarifies the nature of this threat and addresses two possible solutions (named, respectively, ‘resistance’ and ‘accommodation’). It is argued that neither solution is likely to be successful, at least not without risking many other things we value about social decision-making. The result is a somewhat pessimistic conclusion in which we confront the possibility that we are creating decision-making processes that constrain and limit opportunities for human participation.
Article
The Internet of things (IoT) is the network of objects/things that contain electronics, software, sensors, and actuators, which allows these things to connect, interact, and exchange data. The users, sensors, and networks generate huge amounts of data from which governments can develop applications and gain knowledge using Artificial Intelligence (AI) techniques. Thus, IoT and AI can enable the development of valuable services for citizens, businesses, and public agencies, in multiple domains, such as transportation, energy, healthcare, education, and public safety. This guest editorial for the special issue on IoT and AI for Smart Government, identifies the challenges involved in implementing and adopting these technologies in the public sector, and proposes a comprehensive research framework, which includes both IoT and AI elements for smart government transformation. Subsequently, the editorial provides a brief introduction of the six papers in this special issue. Finally, an agenda for future research on IoT and AI for smart government is presented, based on the proposed framework and gaps in existing literature, supported by the papers that were submitted to this special issue. The agenda comprises four directions i.e., conducting domain-specific studies, going beyond adoption studies to examine implementation and evaluation of these technologies, focusing on specific challenges and thus quick wins, and expanding the existing set of research methods and theoretical foundations used.
Article
Driven by ‘success stories’ reported by private sector firms, government agencies have also started adopting various Artificial Intelligence (AI) technologies in diverse domains (e.g. health, taxation, and education); however, extensive research is required in order to exploit the full potential of AI in the public sector, and leverage various AI technologies to address important problems/needs. This paper makes a contribution in this direction: it presents a novel approach, as well as the architecture of an ICT platform supporting it, for the advanced exploitation of a specific AI technology, namely chatbots, in the public sector in order to address a crucial issue: the improvement of communication between government and citizens (which has for long time been problematic). The proposed approach builds on natural language processing, machine learning and data mining technologies, and leverages existing data of various forms (such as documents containing legislation and directives, structured data from government agencies’ operational systems, social media data, etc.), in order to develop a new digital channel of communication between citizens and government. Making use of appropriately structured and semantically annotated data, this channel enables ‘richer’ and more expressive interaction of citizens with government in everyday language, facilitating and advancing both information seeking and conducting of transactions. Compared to existing digital channels, the proposed approach is appropriate for a wider range of citizens’ interactions, with higher levels of complexity, ambiguity and uncertainty. In close co-operation with three Greek government agencies (the Ministry of Finance, a social security organization, and a big local government organization), this approach has been validated through a series of application scenarios.
Article
Advances in artificial intelligence (AI) have attracted great attention from researchers and practitioners and have opened up a broad range of beneficial opportunities for AI usage in the public sector. Against this background, there is an emerging need for a holistic understanding of the range and impact of AI-based applications and associated challenges. However, previous research considers AI applications and challenges only in isolation and fragmentarily. Given the lack of a comprehensive overview of AI-based applications and challenges for the public sector, our conceptual approach analyzes and compiles relevant insights from scientific literature to provide an integrative overview of AI applications and related challenges. Our results suggest 10 AI application areas, describing their value creation and functioning as well as specific public use cases. In addition, we identify four major dimensions of AI challenges. We finally discuss our findings, deriving implications for theory and practice and providing suggestions for future research.
Article
Hidden algorithms drive decisions at major Silicon Valley and Wall Street firms. Thanks to automation, those firms can approve credit, rank websites, and make myriad other decisions instantaneously. But what are the costs of their methods? And what exactly are they doing with their digital profiles of us? Leaks, whistleblowers, and legal disputes have shed new light on corporate surveillance and the automated judgments it enables. Self-serving and reckless behavior is surprisingly common, and easy to hide in code protected by legal and real secrecy. Even after billions of dollars of fines have been levied, underfunded regulators may have only scratched the surface of troublingly monopolistic and exploitative practices. Drawing on the work of social scientists, attorneys, and technologists, The Black Box Society offers a bold new account of the political economy of big data. Data-driven corporations play an ever larger role in determining opportunity and risk. But they depend on automated judgments that may be wrong, biased, or destructive. Their black boxes endanger all of us. Faulty data, invalid assumptions, and defective models can’t be corrected when they are hidden. Frank Pasquale exposes how powerful interests abuse secrecy for profit and explains ways to rein them in. Demanding transparency is only the first step. An intelligible society would assure that key decisions of its most important firms are fair, nondiscriminatory, and open to criticism. Silicon Valley and Wall Street need to accept as much accountability as they impose on others. In this interview with Lawrence Joseph, Frank Pasquale describes the aims and methods of the book.