Recently, the potential role of Artificial Intelligence (AI) technologies both in the public sector and the private sector have focused the attention of the scholarly community. In governmental contexts, previous research has shown different applications of these technologies for different purposes, i.e. data analytics, process automatization or recommendation systems, among others. In this paper we focus on the analysis of one of these public service delivery streams related to citizen engagement and information provision through the application of “chatbots” or “conversational agents” in the public sector. Although the chatbot concept is not new, current development in AI technologies such as machine learning (ML) or natural language processing (NLP) have raised expectations for its adoption in public sector organizations. Some scholars have emphasized the potential of these conversational agents for an 24/7 open administration and evaluated its social characteristics impact on citizen perception and public trust, different chatbot typologies, and specific policy areas of adoption (i.e. public health, education, etc.). However, the absence of a synthesis of these different research approaches and topics impedes a cohesive understanding of the state of the art of this topic, hindering the development of effective strategies and guidelines for future research, implementation, and evaluation of chatbot application in the context of the public sector. To address this research gap, we aim to understand how chatbot application in the context of the public sector have been investigated in previous literature. For this purpose, we have conducted a systematic literature review (SLR) on chatbot application in the context of the public sector through the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) protocol. The final sample includes a total of 69 peer review articles. The main contribution of this study is twofold: (1) to provide an extensive interpretation of research about the topic of chatbots in the public sector, and (2) to identify key issues to improve future research and practice in this field of research.

Healthcare systems are subject to demands far beyond what they can supply, significantly affecting quality, affordability, and accessibility (LUMC, 2018, 2024; Vermeer, 2024). Adopting Artificial Intelligence (AI) technologies in healthcare organizations may increase care quality while lowering costs, improving both patients’ and providers’ experiences, with commensurate increases in public value (Ahern et al., 2011; Criado & Gil-Garcia, 2019; Davenport & Kalakota, 2019; Lee & Yoon, 2021; Shah et al., 2023; Shamszare & Choudhury, 2023). AI adoption in healthcare is conditional on the support of patients, healthcare providers, and society in general (Elahi & Cook, 2023). We identify three lacunae in this literature. First, for AI to be effective in practice, a wide range of stakeholders must support it (Meijer & Thaens, 2021; Moore, 2001; Shaw et al., 2019), but we know too little about whether stakeholders feel the outcomes are worth the costs in situ. Second, the required organizational competencies for AI technologies may conflict with the safeguarding of public values, thereby creating a practical implementation challenge for healthcare organizations (Dingelstad et al., 2022). Third, more knowledge on the goals, objectives, costs, and risks before the adoption phase are necessary (Bailey & Barley, 2020; Young et al., 2021). To address these gaps, we answer the following question: Why do different stakeholder groups in healthcare adopt AI?

By answering this question, we contribute to the public sector AI literature by empirically studying the entire adoption process of an AI tool (Mergel et al., 2023; Orlikowski, 2000). To do this, we conducted a qualitative case study at the radiology department of an academic research hospital in the Netherlands that has adopted AI in the form of new software on its MRI machines designed to make scans quicker and higher quality. Data collection included ethnographic observation (currently 27 days spread over 6 months), (pilot-)interviews with experts and stakeholders (currently 13), document analysis (several internal and external documents), a questionnaire, and field conversations. Data are abductively analyzed with process tracing methods.

Our analysis answers the call for empirical research on AI adoption trajectories that incorporates an extended, social view on the technology before the implementation phase (Bailey & Barley, 2020; Mergel et al., 2023; Orlikowski, 2000; Wirtz et al., 2022). We make a contribution to innovation literature by taking a holistic approach to study the AI adoption process, as opposed to studying the pre-adoption, adoption, and post-adoption separately. Our findings illuminate the intended values of AI adoption according to involved stakeholders, as well as their perspective on the feasibility of intended outcomes. We make a further contribution to public management by aggregating these perspectives to the organizational dynamics at the department level.

AI strategies from and in public administrations have been revealed as an indicator of states’ priorities, ethical values, relationships with the private sector or even international cooperation. Nonetheless, existing literature about AI strategies in government is neither uniform nor comparative, which hinders international approaches. This paper aims to fill this gap at the time that it provides an original frame for analysis by establishing connections between categories of analysis of AI strategies in government from existing academic publications. We set three research questions: RQ1: What are the main features of the academic literature about AI strategies in government? RQ2: How is the clustering of themes in publications based on AI strategies in government? And, RQ3: Which categories can be identified in classifications within publications on AI strategies and how do they interact with clusters? Departing from a systematic literature review based on PRISMA methodology, we search different patterns in AI strategies in government used according to their key dimensions and examine the interrelations among them by studying coding segments. We expect that our study will facilitate understanding both the different categories used and the interconnections among them, enabling their examination and consistency for AI strategies in government. Besides, we will assemble a framework to standardize the different category patterns of AI strategies. This draws a first unified approach to facilitate comparative analysis of national AI policies and strategies. Also, it suggests some ideas for practitioners to understand and better design AI strategies, promoting an international dialogue about the implications of their decisions in this area.

The public sector, with its diverse range of organisations and services, plays a critical role in our economy and society. Public Management as it has been shaped in the Post-New Public Management era, attempts to deliver efficiency, and effectiveness but at the same time respect the distinctive differences between the private and public sector. The technological advancements particularly the introduction of digitalisation and very recently Artificial Intelligence, have transformed the public workplace. The proponents of digitalisation build their case on the efficiency, effectiveness, and improvement of work-life balance for employees. The critics argue that it serves as a substitute for actual skills and knowledge that are required on the job and empowers managers to reduce the resistance and the number of jobs in the workplace.

This paper explores how digitalization has affected Employee Relations in the public sector. It will also investigate the role of digitalization in delivering the objectives of Post-New Public Management. Finally, it will try to devise the relationship between public management, employee relation theories, and digitalization, and examine whether the Labour Process Theory can provide a tool to understand this relationship.