Navigating Governmental Choices: A Comprehensive Review of Artificial Intelligence’s Impact on Decision-Making

Abstract

The integration of artificial intelligence (AI) into government decision-making is rapidly gaining traction in public administration and politics. This scoping review, guided by PRISMA protocols, examines 50 articles from reputable sources like Scopus and SpringerLink to analyze the trends, benefits, and challenges of AI in governance. While AI offers substantial potential to enhance government efficiency and service delivery, significant barriers remain, including concerns about bias, transparency, public acceptance, and accountability. This review underscores the need for ongoing research and dialogue on the ethical, social, and practical implications of AI in government to ensure the responsible and inclusive adoption of AI-driven public services.

1. Introduction

In recent years, there has been a transformative shift in the landscape of government management, moving away from traditional paradigms towards a more comprehensive and diverse approach. This departure from conventional systems is characterized by a meticulous analysis of various facets requiring updates to ensure the effective provision of public services [1]. The resulting changes, while necessitating increased investment, are aimed at achieving heightened efficiency and effectiveness in administration, leveraging available financial resources. One pivotal driver of this transformation is the integration of technology, particularly artificial intelligence (AI), which not only facilitates timely decision-making but also challenges both governmental bodies and the public to assimilate the evolving reality. The application of AI in public services involves the deployment of automation techniques, streamlining routine tasks that are cumbersome for human processing, and freeing up human resources for critical decision-making. This paradigm shift enhances administrative efficiency across sectors such as healthcare, agriculture, security, education, and food sovereignty.

The comprehensive and diverse approach to proper government management has led to the breaking of mental paradigms from traditional theories applied in countries seeking to depart from the conventional public system. This involves a detailed analysis of areas requiring updates for the proper provision of public services. Unfortunately, these changes necessitate increased investment and aim to achieve efficiency and effectiveness in administration with the available financial resources [2].

Technology enables timely and necessary action to make decisions in each innovation that entities will implement. With the development of artificial intelligence, it poses a challenge for both governments and the public in assimilating the reality we face [3]. Its objective is to meet the unmet basic needs concerning the provision of public services, which is the responsibility of central governments.

Artificial intelligence brings forth numerous benefits in the realm of public services. By automating time-consuming and repetitive tasks, AI allows for rapid and accurate execution, empowering human workers to focus on decision-making based on the generated results [4]. Furthermore, AI’s capacity to identify patterns and trends within data is instrumental in predicting and preventing issues before they escalate, particularly in areas such as natural disaster management, national security, and public health. Traditional methods often demand a considerable amount of time, and the swiftness of AI-driven analyses enables a more proactive approach. This strategic redirection of resources towards high-level matters ensures more efficient governance and fosters transparency and traceability, thereby strengthening public trust and accountability.

With the application of artificial intelligence in public services, automation techniques are deployed to handle routine tasks that are difficult for human hands to process manually in large amounts of data. Using appropriate techniques, time-consuming and tedious tasks can be performed quickly and accurately, without disregarding the human workforce, which can then be dedicated to decision-making based on the results, thereby improving administrative efficiency in sectors such as healthcare, agriculture, security, education, and food sovereignty [5].

Another significant aspect is that AI identifies patterns and trends in data that can help predict and prevent problems before they occur [6]. This is valuable in areas such as natural disaster management, national security, and public health, where traditional methods would take a considerable amount of time. This allows resources to be directed towards strategic and high-level matters, and the results obtained can be made more transparent, making processes clearer and more traceable, thus strengthening public trust and enhancing accountability.

The availability of information on the subject under study is significant and continues to grow, as many governments have published reports and policies related to the adoption of AI in their operations and decision-making [7]. These documents provide valuable insights into the strategies, challenges, and approaches implemented in the integration of AI. With the entire bibliography at hand, the aim of this review was to identify recent trends and advancements in the application of artificial intelligence in the design and implementation of public policies. Furthermore, this review sought to assess AI’s impact on government decision-making, benefiting a wide range of stakeholders, from governments and citizens to the academic sector. These studies contribute to the development of more effective policies, transparent systems, and the improved delivery of public services.

Consequently, the application of artificial intelligence to government decision-making has the potential to enhance efficiency, transparency, and the quality of public services, but it also presents significant challenges that must be responsibly addressed. Therefore, research in this field is essential to fully comprehend its implications and to maximize the benefits while mitigating the associated risks [8].

The primary goal of this article is to conduct a comprehensive literature review utilizing the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology. Building upon the insights provided in the preceding paragraphs, the review aims to systematically analyze and synthesize existing research on the integration of artificial intelligence (AI) in government decision-making and public service provision. By adhering to the PRISMA methodology, this review ensures a rigorous and transparent process, incorporating a detailed analysis of relevant studies, reports, and policies published by various governments. Through this systematic approach, the article seeks to identify and highlight recent trends, advancements, challenges, and impacts associated with the application of AI in the design and implementation of public policies. Furthermore, it aims to contribute valuable insights that can benefit governments, citizens, and the academic sector, fostering the development of more effective policies, transparent systems, and improved public service delivery.

Through this exploration, we aim to shed light on the potential benefits and challenges of AI integration, fostering discussions that can shape more effective policies, transparent systems, and improved public service delivery. The highlights of the review are as follows:

• Methodological precision through PRISMA: The paper aims to uphold methodological rigor by employing the PRISMA methodology, ensuring a systematic and transparent approach to the literature review. This methodology guided the selection, screening, and inclusion of relevant studies, reports, and policies related to the integration of artificial intelligence (AI) in government decision-making and public service provision.
• Synthesis of insights and trends: The primary goal was to systematically analyze and synthesize existing research to identify recent trends, advancements, challenges, and impacts associated with the application of AI in the formulation and execution of public policies. By doing so, the paper aims to provide a comprehensive overview of the current state of knowledge in this domain, offering valuable insights for policymakers, researchers, and stakeholders.
• Contribution to effective governance: Ultimately, the review seeks to contribute to the development of more effective policies, transparent systems, and improved public service delivery. By distilling key findings and lessons from the literature, the paper aims to inform governmental bodies, citizens, and the academic sector, fostering a better understanding of the implications of AI in government decision-making and enhancing the overall quality of public services.

The present research work consists of five sections. Section 1 contains the introduction. Section 2 presents the methodology. The subsequent Section 3 displays the results. Section 4 addresses and discusses the questions posed in the research methodology, and finally, Section 5 presents the conclusions.

2. Methodology

This study involved conducting a comprehensive literature review focusing on the intersection of artificial intelligence (AI) and governmental decision-making. To ensure methodological rigor, the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology was employed. Recognized for its importance in systematic reviews, PRISMA emphasizes standardization, transparency, bias minimization, and quality enhancement, contributing to the credibility and utility of the review within the scientific community.

In adherence to the principles of open and transparent science, we have pre-registered our study protocol on the Open Science Framework (OSF). The complete methodology, including detailed hypotheses, study design, data collection procedures, sample size justification, and planned statistical analyses, is publicly available at https://osf.io/bsgx5 (accessed on 10 May 2024). This pre-registration serves as a transparent roadmap for our research, enabling an independent evaluation of our methodological approach and helping mitigate potential biases in the research process. The OSF registry provides comprehensive information about our manipulated and measured variables, as well as the specific statistical models we intend to employ for hypothesis testing. We encourage readers to review this pre-registration for a thorough understanding of our approach. In the interest of full transparency, any deviations from this pre-registered plan that may occur during the course of the study will be explicitly noted and justified in our final analysis and reporting. This commitment to open science practices underscores our dedication to conducting robust, replicable research on the critical topic of artificial intelligence in government decision-making.

To advance the research, various reputable sources, including Scopus, were accessed to gather substantial information on the subject. Scopus underscores the significance of artificial intelligence in the contemporary economy, highlighting challenges faced by traditional models [9]. In addition to this, the bibliographic–documentary research methodology was utilized, conducting an analysis of previously researched information and data pertaining to AI and government decision-making. This analysis delves into the innovative techniques adopted by countries in this domain. A systematic approach was established for the literature review, involving the identification, interpretation, and evaluation of existing information. This process commenced with the use of carefully chosen keywords to facilitate a targeted search.

The search criteria were refined by segmenting the publication timeline, relevance to the research questions, and publications that reference these citations. By employing keywords such as “artificial intelligence” and “governmental decisions”, pertinent bibliographic information was uncovered [10]. Through this methodological framework, the aim was to accurately summarize all collected bibliographic data, providing a guide for assessing the quality of the systematic review. Given this context, the present review concentrated on extracting information from diverse and esteemed databases accessible to the authors. These include renowned platforms such as SpringerLink, MDPI, Scopus, Web of Science (WoS), ACM, and IEEEXplore. It is essential to note that, although Google Scholar, Scopus, and WoS stand out as widely utilized academic tools, Google Scholar was excluded from consideration due to its inclusion of theses and reports that do not align with the selection criteria for this research. Additionally, the review followed a structured approach comprising three key steps: (i) defining research questions, (ii) conducting a meticulous document search, and (iii) systematically selecting relevant papers.

2.1. Review Protocol

In accordance with academic studies published in reputable journals, a comprehensive array of scientific article search tools was identified to enhance our understanding of the subject matter. A meticulous review of these tools was conducted, prioritizing those that make substantial contributions to ensure the quality and transparency of the review process. This rigorous selection process aimed to furnish valuable insights for informed decision-making within the governmental context and pinpoint areas warranting future research endeavors.

Once the topic under investigation was clearly defined, we meticulously curated the final bibliography, ensuring that it aligned with the research focus. This curated bibliography served as a foundational resource for an in-depth understanding and documentation in subsequent stages of the review.

Following the review of the bibliography, a nuanced assessment became imperative. It was crucial to establish a clear focus for examination, aligning it with the study’s objectives. This focused approach guided the collection of data and information, ensuring a harmonious integration with the overarching goals of the research.

2.2. Research Questions

Each research question was carefully formulated in alignment with the central issue under consideration. The objective was that, upon concluding this research endeavor, each question would be addressed with clarity and precision. This endeavor was positioned to make a positive impact on the economic and political development of our nation. The current state of underdevelopment at the national level underscores the critical need for a thorough examination, as there is a discernible absence of research conducted by Ecuadorian authors dedicated to this crucial issue.

The multifaceted inquiry into how artificial intelligence influences governmental decision-making, particularly in enhancing public confidence, serves as a pivotal focus of this study. The research aims to discern how AI, through improved efficiency and effectiveness, contributes to governmental decision-making by expediting data analysis, identifying patterns, and furnishing recommendations grounded in accurate, real-time information. This contribution, in turn, fortifies a government’s capacity to wield this technological tool for the advancement of the country.

Additionally, a detailed exploration is intended to illuminate the evolution of policies and regulations in response to the escalating integration of artificial intelligence in governmental decision-making, elucidating their practical impact. This nuanced investigation seeks to enhance the management of AI, ensuring the acquisition of more precise and accurate data. The research structure encompasses four core questions, strategically designed to encapsulate the spectrum of artificial intelligence’s role in governmental decision-making (

Table 1. Formulation of research questions.

Code	Research Question	Aim
RQ1	How does the integration of artificial intelligence in governmental decision-making enhance efficiency and effectiveness in public administration?	To examine the specific ways in which AI improves government operations, including data processing capabilities, service delivery, and policy formulation.
RQ2	What are the recent trends and advancements in the application of artificial intelligence to the design and implementation of public policies?	To identify the latest innovations and approaches in the use of artificial intelligence in the field of public policies, assessing the current state of research and policy implementation based on AI.
RQ3	How have policies and regulations been developed in different countries to oversee and regulate the use of artificial intelligence in governmental decision-making?	To analyze the regulatory frameworks being established globally to govern AI use in public administration, identifying best practices and potential areas for improvement.
RQ4	What are the primary challenges and ethical considerations in implementing artificial intelligence systems for government decision-making?	To explore the technical, social, and ethical obstacles faced when integrating AI into public sector operations, including issues of transparency, accountability, and potential biases.

).

AI systems can streamline administrative processes and enhance the efficiency of delivering public services. This can lead to increased accessibility and swiftness in addressing the needs of citizens, which, in turn, can boost satisfaction.

2.3. Document Search

This section involves a meticulous search for relevant documents across esteemed academic databases, aligning with the defined research questions. The PRISMA methodology, known for its emphasis on standardization and quality enhancement, provided a robust foundation for this document search, contributing to the rigor and credibility of our systematic review. This sub-section is integral to the overall research process, serving as a critical step in identifying, screening, and selecting pertinent studies that contributed substantively to our understanding of the intersection between artificial intelligence and governmental decision-making.

List of Databases Used

In the pursuit of comprehensive and current insights into the subject under study, information was diligently gathered from various database platforms, including Scopus, Google Scholar, Web of Science, and others. The temporal scope of this database search was specifically focused on the last five years, spanning from 2019 to 2023. This strategic timeframe ensured that the research drew upon the most recent and relevant literature, capturing developments and advancements within the dynamic landscape of artificial intelligence and governmental decision-making. By limiting the search to this five-year window, the aim was to provide a nuanced understanding that reflects the contemporary state of the field, acknowledging the rapid evolution of technology and its applications within the specified timeframe. This temporal specificity enhances the relevance and timeliness of the gathered information, aligning with the overarching goal of conducting a thorough and up-to-date literature review. See

Table 2. Search terms and justification of research questions.

RQ	Key Concepts	Search Terms	Justification
RQ1	- Artificial intelligence - Government - Efficiency - Effectiveness	- “artificial intelligence“, “machine learning“, “deep learning“ - “government“, “public administration“, “public sector“ - “efficiency“, “effectiveness“, “optimization“	These terms capture studies discussing AI applications in government that lead to improved operations and service delivery. The combination allows for the identification of research on AI’s impact on governmental efficiency.
RQ2	- Recent developments - AI applications - Public policy	- “trends“, “advancements“, “innovations“ - “artificial intelligence“, “machine learning“ - “public policy“, “policy design“, “policy implementation“	This combination of terms, coupled with the date range filter (2019–2023), ensured the capture of the most recent innovations in AI applications for public policy.
RQ3	- AI regulation - Policy development - Governmental oversight	- “regulation“, “policy“, “governance“ - “artificial intelligence“, “AI“ - “government“, “public sector“	These terms were designed to identify studies discussing the development of regulatory frameworks for AI in government across different countries.
RQ4	- Challenges - Ethical considerations - AI implementation	- “challenges“, “barriers“, “obstacles“ - “ethics“, “ethical considerations“, “moral implications“ - “artificial intelligence“, “AI systems“, “machine learning“	This set of terms allows for the identification of studies discussing both technical and ethical challenges in implementing AI in governmental contexts.

.

Prior to initiating the exploration of studies delineating the influence of artificial intelligence on governmental decision-making, a methodically structured and succinct search design was adopted. This preliminary phase aimed at ensuring the availability of research papers directly pertinent to our subject of study. The employment of connectors, including AND, OR, and NOT, along with judiciously applied filters, played a crucial role in streamlining the search process. This strategic approach not only enhanced the precision of our search but also contributed to the enrichment of the bibliography with studies most germane to our research objectives.

Our search strategy was meticulously crafted to align with the research questions and encompass the full spectrum of the literature on artificial intelligence in government decision-making. We employed a comprehensive set of key terms, strategically chosen to capture the multifaceted nature of this field. The core of our search revolved around artificial intelligence-related terms such as “artificial intelligence”, “machine learning”, and “deep learning”. These terms were selected to ensure coverage of the wide array of AI technologies relevant to governmental applications, from basic automation to sophisticated predictive models.

To focus our search on the governmental context, we incorporated terms like “government”, “public administration”, and “public sector”. This approach allowed us to filter out studies that might discuss AI applications in other domains, ensuring that our review remained centered on the public sector. Additionally, we included decision-making related terms such as “decision making”, “policy making”, and “governance” to align with our research focus on how AI impacts governmental decision processes. This combination of terms enabled us to capture studies that specifically addressed the intersection of AI technologies and governmental decision-making mechanisms.

Our search query was carefully designed to address each of our research questions comprehensively. For RQ1, which focuses on how AI enhances efficiency and effectiveness in government, we combined AI terms with government and decision-making terms. This approach allowed us to identify studies discussing the practical improvements AI brings to governmental operations and service delivery. RQ2, concerned with recent trends and advancements, was addressed by limiting our search to the years 2019–2023 and including broad AI terms. This strategy ensured that we captured the latest innovations and approaches in AI applications to public policy, providing a current snapshot of the field.

To address RQ3, which examines policy and regulation development, we included terms like “policy” and “governance” in our search. This inclusion helped us identify studies discussing the evolving regulatory frameworks for AI in government across different jurisdictions. Finally, for RQ4, which explores challenges and ethical considerations, our broad search terms allowed for the inclusion of studies discussing both technical and ethical challenges of AI implementation in government. This comprehensive approach ensured that our review captured not only the potential benefits of AI in government but also the critical discussions surrounding its responsible and ethical implementation.

Leveraging important document databases facilitates the affirmation that the latest research pertaining to artificial intelligence and governmental decisions is contemporaneous. Notably, numerous authors have contributed to this field within the current year, thereby offering valuable insights and an up-to-date perspective on the research topic, supported by recent bibliographic references.

The systematic search strategy employed in this study was designed to ensure comprehensive coverage of the literature relevant to artificial intelligence in government decision-making. We utilized multiple databases to capture a wide range of academic publications, recognizing that different databases may have varying strengths in coverage across disciplines. This approach allowed us to minimize bias and ensure a thorough representation of the current state of research in this field.

Our search queries were tailored to each database’s specific syntax and capabilities while maintaining consistency in the core concepts being explored. The following table presents the exact search queries used for each of the primary databases consulted in this study. These queries were constructed to balance specificity and sensitivity with the aim of capturing all relevant literature while minimizing the inclusion of irrelevant studies.

The search queries presented in

Table 3. Search queries by database.

Database	Search Query
Scopus	TITLE-ABS-KEY((“artificial intelligence“ or “machine learning“ or “deep learning“) and (“government“ or “public administration“ or “public sector“) and (“decision making“ or “policy making“ or “governance“))
SpringerLink	(“artificial intelligence“ or “machine learning“) and (“government decision-making“ or “public administration“ or “public policy“)
Web of Science	TS=((“artificial intelligence“ or “machine learning“) and (“government“ or “public administration“) and (“decision making“ or “policy“))

were carefully constructed to capture the intersection of artificial intelligence and government decision-making across various contexts. Each query was adapted to the specific syntax requirements of its respective database while maintaining conceptual consistency. This approach ensured that we captured a comprehensive set of the relevant literature, regardless of database-specific indexing practices.

In the Scopus database, we utilized the TITLE-ABS-KEY field code to search within titles, abstracts, and keywords. This broad search scope allowed us to capture articles that may not explicitly mention all terms in their titles but discuss relevant concepts in their abstracts or keywords. For SpringerLink, which has a simpler search interface, we used a more streamlined query that still encompassed the core concepts of our research. The Web of Science query used the TS field tag to search in titles, abstracts, author keywords, and Keywords Plus^®.

These carefully crafted search strategies resulted in an initial pool of studies that comprehensively represented the current state of research on AI in government decision-making. The results from these searches were then subjected to our rigorous screening and eligibility assessment process, as outlined in our PRISMA flow diagram. This meticulous approach ensured that our final selection of studies for review was both comprehensive and highly relevant to our research questions.

2.4. Study Selection

Our meticulous study selection process, grounded in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology, was driven by the recognition that artificial intelligence (AI), beyond serving as a systematic tool, holds the potential to shape strategies for the economic development of a territory. The intersection of politics and technological dependence in the public sphere implies an increasing reliance on technological tools for political decision-making, dissemination, and propaganda.

Within this broader landscape, AI offers the promise of a more nuanced understanding of global issues on a large scale, paving the way for enhanced and alternative solutions. Addressing the pervasive issue of corruption in global politics while acknowledging the absence of a foolproof formula, AI might contribute to transparency by creating oversight mechanisms within public organizations.

In Figure 1, a PRISMA flowchart [11] illustrates our meticulous selection process. Fifty articles underwent rigorous review by our research group, ensuring the meticulous selection of the most relevant ones contributing to our central research topic. This methodical application of the PRISMA methodology added a layer of rigor and transparency to our study, aligning with best practices in systematic reviews and enhancing the reliability of our findings.

The PRISMA flow chart generated from our systematic review process provides a comprehensive visual representation of our literature search and selection methodology. In the identification phase, we initially identified 1243 records through database searches, specifically 543 from Scopus, 412 from SpringerLink, and 288 from Web of Science. This diverse range of databases ensured a broad coverage of the literature on artificial intelligence in government decision-making. We supplemented this with 10 additional records identified through other methods: 1 from website searches and 9 from citation searching, demonstrating our commitment to a thorough and multifaceted search strategy.

The identification of new studies via other methods, while yielding a smaller number of records compared to database searches, played a crucial role in ensuring the comprehensiveness of our review. This approach allowed us to capture relevant literature that might not have been indexed in the primary databases, including recent publications, gray literature, or studies from interdisciplinary sources. The inclusion of these additional sources enhanced the robustness of our review by mitigating potential biases inherent to relying solely on traditional database searches.

During the screening phase, we first removed 213 duplicate records, leaving 1030 unique records for initial screening. This step is crucial for maintaining the integrity of the review process and avoiding redundant evaluations. Of these, 800 records were excluded based on title and abstract screening, leaving 230 full-text articles to be assessed for eligibility. This substantial reduction demonstrates the rigorous application of our inclusion and exclusion criteria, ensuring that only the most relevant studies progressed to full-text review.

The full-text assessment stage resulted in the exclusion of 187 articles, with reasons categorized into three main groups: 73 for not focusing specifically on government decision-making (Reason1), 52 for lacking empirical data or being purely theoretical (Reason2), and 62 for not addressing AI applications directly (Reason3). This detailed breakdown of exclusion reasons provides transparency in our decision-making process and allows readers to understand the specific criteria applied in our final selection. It is worth noting that all 10 records identified through other methods were assessed, with 9 excluded for not focusing on government decision-making and 1 for lacking empirical data, highlighting the rigorous evaluation applied to all sources, regardless of their origin.

Ultimately, our systematic review process culminated in the inclusion of 43 studies, representing the most relevant and high-quality research addressing our research questions on artificial intelligence in government decision-making. This final selection reflects a careful balance between comprehensiveness and specificity, ensuring that our review captures the current state of knowledge in the field while maintaining a focused and manageable scope.

Our search queries were meticulously crafted to balance sensitivity and specificity, using a combination of controlled vocabulary and free-text terms relevant to artificial intelligence and government decision-making. We tailored these queries to each database’s specific syntax and capabilities while maintaining conceptual consistency across all searches. The data selection process involved a multi-stage screening approach, with clearly defined inclusion and exclusion criteria applied at each stage. The initial screening of titles and abstracts was conducted independently by two reviewers, with disagreements resolved through discussion or by a third reviewer. The full-text assessment similarly involved a dual review, with a standardized form used to record reasons for exclusion.

Our data analysis process involved both quantitative and qualitative elements. We extracted predefined data from the included studies using a standardized form, capturing key information on study characteristics, methodologies, and findings. Thematic analysis was then employed to synthesize the findings, identifying recurring themes and patterns across the included studies. This approach allowed us to generate meaningful insights while maintaining the nuance and complexity inherent to the diverse body of literature on AI in government decision-making.

2.4.1. Quality Assessment and Risk of Bias Evaluation

In the pursuit of methodological rigor and the credibility of our systematic review, the ensuing subsection delves into the crucial aspects of quality assessment and the risk of bias evaluation. This stage is pivotal in ensuring the reliability and robustness of the selected studies, as we systematically evaluated their quality and potential biases. Adhering to established criteria and employing transparent methodologies, this assessment aimed to discern the internal validity of each study, providing a comprehensive understanding of the strengths and limitations inherent to the body of literature under review. This meticulous evaluation process enhanced the trustworthiness of our findings and contributed to the overarching goal of providing a nuanced and well-founded perspective on the intersection of artificial intelligence and governmental decision-making.

2.4.2. Methods Used to Assess the Quality of the Included Studies

The comprehensive evaluation of methodological and informational quality constitutes a pivotal step in scrutinizing the conduct and application of research, particularly in the realm of how artificial intelligence influences governmental decision-making. The objective is to establish a robust connection by meticulously assessing the methodological quality of the cited scientific articles. This assessment is facilitated through the utilization of a checklist specifically designed for evaluating information quality in research within the economic, political, and social fields. The criteria encompass basic characteristics related to the number of items, validity and reliability, applicability, and limitations.

To ensure a thorough examination, each study was scrutinized against these criteria, providing a nuanced understanding of its methodological soundness and the reliability of the information that it contributes. The emphasis on validity and reliability ensured that the research under review met stringent scientific standards, bolstering the credibility of our systematic review. Applicability and limitations were also carefully considered, offering insights into the practical implications of the findings and acknowledging any inherent constraints. This methodological scrutiny enhanced the discernment of the overall quality of the included studies, contributing to the robustness of our systematic review.

The evaluation criteria employed in this subsection harmoniously converged with the broader objectives of our systematic review. The assessment of methodological quality ensured that the selected studies upheld rigorous scientific standards, fostering confidence in the reliability and validity of the insights derived from them. By integrating a checklist tailored to the economic, political, and social contexts, we aligned the evaluation process with the specific nuances of our research domain. This meticulous integration allowed for a nuanced understanding of the methodological rigor of each study, contributing to the overall integrity of our systematic review.

Moreover, the evaluation process served to highlight the relevance of the selected studies to the economic, political, and social aspects of governmental decision-making influenced by artificial intelligence. By systematically applying the checklist, we discerned not only the strengths of the research but also its limitations and applicability. This holistic approach ensured that our systematic review went beyond a mere enumeration of findings, offering a critical appraisal that enhances the scholarly value of the included studies.

2.4.3. Consideration of Bias Risk in the Included Studies

In addition to assessing methodological quality, our evaluation process diligently considered potential biases inherent to the selected studies. Recognizing that bias can impact the objectivity and generalizability of research findings, we systematically examined each study for indications of bias. This included a critical assessment of factors such as sampling bias, publication bias, and researcher bias. By conscientiously considering these elements, our systematic review aimed to provide a balanced and objective synthesis of the existing literature, acknowledging and mitigating potential sources of bias to bolster the credibility of our findings.

To sort and analyze the gathered information, the following criteria detailed in

Table 4. Criteria guiding the selection of studies in the systematic review, emphasizing aspects such as publication type, research theme, geographical scope, and methodology.

Inclusion Criteria	Description
Publication type	Peer-reviewed journal articles and conference papers, government reports, and books.
Research theme	Studies had to focus on the application of artificial intelligence in governmental decision-making.
Geographic scope	No geographical restrictions; studies from local, regional, national, and international levels were included.
Methodology	Studies that present clear methodologies and approaches in the application of artificial intelligence in governmental decision-making.

were considered:

The criteria for analyzing the collected information were as follows: Selection was based on a meticulous evaluation of publication types, limiting the inclusion to peer-reviewed journal articles, conference papers, government reports, and books. The research theme centered on the application of artificial intelligence to governmental decision-making, emphasizing studies’ relevance. The geographical scope imposed no restrictions, encompassing studies from local to international levels. Methodology was a critical criterion, ensuring that the selected studies present clear methodologies and approaches in applying artificial intelligence to governmental decision-making.

For prioritization, considerations included relevance to the topic, recentness, language usage (Spanish and English), and methodologies aligned with expected results. Additionally, a thorough review ensured that references aligned with the information presented in each paragraph. Systematic biases in participant sampling, occurring during selection and data collection, can impact sample representativeness, affecting validity and generalizability. The research sampling cycle denotes instances when a sample was not randomly selected or lacked careful consideration to ensure that all population elements had an equal probability of inclusion.

3. Results

In this section, a summary of the works selected from the review source is provided as follows:

Technology is causing a disruption at an unprecedented speed and magnitude in history. Technological innovations, large-scale data analytics, machine learning, and artificial intelligence are bringing about profound changes in our lives, both personally and professionally. We are entering a context where many current occupations will disappear, while new ones will emerge, demanding different skill sets. Government officials are unprepared for the challenges they must encounter in the face of this rapid and disruptive change, which is not gradual. Many of the governmental structures and processes that have evolved over the past centuries will likely become irrelevant in the near future. There is an urgent need to lay the groundwork for governments to reconsider how they can provide more efficient services [12].

Ref. [13] provided evidence that the implementation of this mechanism is on the rise in the public sector. However, there are shortcomings in understanding and acceptability among the public, which depend on various factors. One of the main factors is the perception of utility. In other words, if the population perceives that this application enhances the efficiency and effectiveness of public services by making responses to their needs quicker and simplifying existing bureaucratic processes, the public’s response will be swifter. This is because they will agree that AI can optimize various resources.

Although there is a growing consensus about the potential of analytical and cognitive artificial intelligence (AI) tools to positively transform government, it is also evident that AI poses challenges to traditional government decision-making processes and threatens the democratic values in which they are framed. These conditions support cautious approaches to AI that focus on cultivating and maintaining public trust. There are tools that illustrate the differences between policy analysis and decision-making, as traditionally practiced, and how they are evolving in the current context of AI, along with the challenges this poses for using reliable AI. Ref. [10] provided a set of recommendations for practices, processes, and governance structures in government that foster trust in AI and suggested lines of research to support them.

Several countries have implemented national policies and guidelines to balance the inherent benefits and challenges of innovative technologies. The aim of [14] was to identify the most prominent issues addressed in national artificial intelligence (AI) policies and assess their relative relevance across different countries. This study combined the results of a thematic modeling analysis of 30 national AI policies with a qualitative content analysis of policy documents. As a result of the analysis conducted, it identified fourteen common key themes in national AI policies, focusing on areas related to education, technology, and governance ethics and legality, and social welfare was then examined. The authors analyzed the coexistence of these issues in all countries to determine the degree of emphasis on each theme in each nation.

AI plays a fundamental role in the public domain, from understanding societal needs to managing traffic flow in both large and small cities, as well as overseeing public transportation systems and assisting the police in data management and citizen-government communication. Ref. [15] argued that, in the context of a slowing global economy, the development of AI could be crucial in boosting labor productivity, stimulating gross domestic product (GDP) growth and enhancing federal, regional, and local communication. This creates opportunities for new digital business strategies. To pursue this initiative, reliable measurement tools aligned with each nation’s strategic AI goals are essential.

Ethics, applicability, responsibility, and accountability are crucial concepts for addressing the social implications of artificial intelligence and machine learning (AI), according to [16]. However, they are insufficient to guide public entities in effectively regulating and implementing AI. Recent designs for AI governance have contributed to its implementation by identifying the processes and governance levels where they should be considered, but they do not provide clear guidance on their implementation and understanding by public sector professionals.

The results of the research conducted by [17] are relevant for local authorities, as they indicate that smart cities must incorporate social innovation into their decision-making processes. This is because smart decision-making involves sharing data collected with entrepreneurs, businesses, and industries, benefiting society as a whole and all relevant stakeholders, including actors in social innovation.

As the pressure to implement automated decision-making systems in the public sector grows, the article by [18] proposed exploring how the use of AI in government, in relation to existing data governance frameworks and national regulatory practices, may be exacerbating preexisting balances. To this end, they investigated legal and policy instruments related to the use of AI to strengthen the control over immigration processes in various countries to personalize the experience of digital services, taking as an example other countries that have already implemented this. They analyzed the specific effects of automated decision support systems in public services and the increasing expectations that governments play a more prominent role in digital society, ensuring that technological potential is achieved while negative effects are controlled and minimized.

It is widely acknowledged that artificial intelligence (AI) has the potential to significantly enhance the management of internal activities and the delivery of services in the public sector. However, realizing this potential depends on the effective implementation of the technology, influenced by unique factors that can either facilitate or hinder its adoption. Despite its importance, there is still a limited understanding of these factors and how they influence the implementation of AI. This has led to [19] investigating various contributions that empirically and practically enrich the existing knowledge on these subjects.

Research conducted by [20] stemmed from the reality of applying artificial intelligence (AI) in the public sector, generating knowledge from real-world experiences. The cited article aimed to analyze the obstacles that government organizations face when adopting AI. It was based on eight comprehensive case studies of AI solutions.

The public sector is moving towards the digital era, and it aims to capitalize on past innovations in information technology (IT). A more systematic approach is needed to anticipate the strategic potential of IT and promote ongoing innovation. This digital transformation impacts every aspect of the public sector, from data collection and analysis to service delivery. A clear example of this is evident in the statements made by [21], discussing big data that enables faster, smarter, personalized government services; with AI automating and optimizing government functions. Furthermore, it examined how the public sector leverages IT innovations to maintain productivity, enhance cross-functional collaboration, and improve the delivery of personalized e-services.

For ref. [22], the aim of the article was to enrich the theoretical discussion concerning the relationships between digital governance and social innovation and their impact on policy formulation to generate and harness the value of effective solutions addressing social challenges. It encompassed the strengths and obstacles involved, with particular attention to the increasing number of national strategies focused on innovation supported via artificial intelligence (AI) and the resulting influx of investments in this field. To achieve this, the article presented a conceptual framework that sought to connect, on the one hand, the fundamental aspects and value drivers in digital governance for social innovation and methods derived from the theory of complex systems for policy development. On the other hand, it aimed to position national AI initiatives in relation to each nation’s welfare initiatives.

Ref. [23] addressed perceived exclusivity and paternalism by setting goals and standards in the context of explainable artificial intelligence (AI). It also explored implications for public AI governance. The authors argued that the increasing use of AI-based decisions, including the development of autonomous systems, not only poses risks to human autonomy but also defines standards characterized by their lack of openness to effective public participation. As several countries progress in AI governance, one of the essential tasks lies in ensuring not only the technical “explainability“ of AI systems but also the questionability of relevant standards and regulations, as well as the openness of government institutions and processes to relevant accountability in each entity, year by year.

The technologies of artificial intelligence (AI) in public administration are gaining increasing attention due to the potential benefits they can offer in improving government operations. However, there are still difficulties in translating these technological opportunities into tangible public value for government institutions. One of the factors hindering this progress is the lack of AI capacity within public organizations. This was analyzed by [20], who identified various essential elements in the development and successful utilization of AI technologies, including tangible, intangible, and human-related aspects. Among the most significant challenges are the capacity for AI development and implementation, the absence of internal technical expertise to maintain and update AI systems, legal barriers to implementing developed systems, and the ability to incorporate organizational changes to ensure that the system remains functional and is used by relevant end users. These factors are critical obstacles to the long-term adoption of AI by public administrations. It was emphasized that both technical and non-technical human skills are crucial for the successful implementation of AI in the field of public administration.

Ref. [24] investigated the impact of introducing artificial intelligence in public entities at a micro level, which implies examining how it influences the roles, competencies, and responsibilities of the individuals involved, the focus of organizational design theory, and a specific AI solution. In this case, the article mentioned how a chatbot could be implemented in a customer service department to collect data, which is one of the common challenges faced by organizations. The results confirmed that the implementation of AI is a highly complex organizational challenge and suggest that teams operate similarly to institutional personnel.

In several already developed countries, the use of artificial intelligence (AI) is prevalent. Machine learning (ML) has become crucial for addressing data management and protection in the context of small and medium-sized enterprises (SMEs). These SMEs have established their own AI- and ML-based cybersecurity strategies, which are assessed daily in their operations management and threat identification. The effectiveness of these strategies depends on the support structure of each country. Ref. [25] employed a methodology involving quantitative and qualitative survey questionnaires directed at SME senior management and both technical and non-technical professionals. The results suggested that SMEs have the appropriate cybersecurity tools, although they may not always be aware of their potential.

Artificial intelligence (AI) has garnered significant interest in society, and it is expected that these technologies will offer significant benefits to public administrations if adopted. Ref. [22] conducted an initial analysis aimed at identifying, categorizing, and understanding current implementations of artificial intelligence in government services. This was done through a documentary research approach, examining available documents describing AI-related projects. The study identified and reviewed 85 AI applications in the public sector of selected European countries. The results provided insights into various landscapes and laid the groundwork for more in-depth research and the formulation of future policy recommendations.

The United Nations (UN) has set forth 17 “Sustainable Development Goals” (SDGs), one of which is SDG 11, focused on developing sustainable cities and communities. In this context, local governments face the challenge of aligning themselves with this goal, and, as a result, they are expanding their efforts to engage citizens in policy formulation and strategy development, often by listening to the public through social media. In the same context, ref. [26] identified that the data analysis process comprises three primary procedures: (1) participation analysis, (2) trend-based analysis, and (3) data collection. The results obtained indicate the following: (1) The COVID-19 pandemic has negatively impacted user participation related to SDG 11. (2) New technologies such as artificial intelligence (AI) are gaining more relevance in assisting cities in achieving SDG 11. And (3) there is an interconnection between the different SDGs, such that progress in one of these goals can influence the advancement of other SDGs.

The application of AI to public administration encompasses two key aspects [27]. Firstly, it enhances the efficiency of administrative machinery. By expanding the capabilities of a public administration to resolve various problems, AI implementation increases the efficiency of the public sector [28]. This can be observed through the establishment of virtual agents in different policy domains such as healthcare and immigration, security control and monitoring utilizing facial recognition to identify criminals, autonomous vehicles for public transportation, chatbots for public service delivery, and image diagnosis to expedite medical care. In all of these scenarios, AI alters the cost–benefit ratio within the public sector, ultimately boosting the efficiency of public administration across various societal sectors [29].

Furthermore, digital tools are transforming both the policy and organizational aspects of public administration [30]. AI is reshaping administrative culture and modifying the ideologies inherent to public management [31]. While sometimes reaffirming traditional values like meritocracy and political neutrality, it also introduces new values such as efficiency and control. Digital transformation within public organizations goes beyond simply reducing the financial costs of public services; it fundamentally changes the institutional functioning of public administration, driven by concerns related to control, costs, convenience, and connectivity.

This disruptive organizational process is precisely what digital governance must address to effectively harness the potential of digital technologies for governance purposes. The disruptive potential offered by digital tools is immense. Digital transformation represents a significant phenomenon for research on public administration, as well as for practitioners. It views technology as a component of a complex innovation process involving organizations, institutions, citizens, and businesses, all working together to transform the value chain of public administration.

A crucial aspect to consider is the adoption of algorithms that fuel the increasing automation and learning within public administration [32]. Algorithms, sequences of instructions for problem-solving or task execution, have been employed in public administration processes for decades to aid decision-making. However, the major disruption today stems from big data analytics, enabling AI systems to learn from massive datasets and make autonomous governance decisions. Machine learning algorithms have been developed to address complex issues by leveraging large volumes of data, and they can operate under supervised or unsupervised learning models to generate predictions for the problems at hand.

Ref. [33] focused on examining the current and emerging applications of AI that will have an impact on most, if not all, functions of human resource management and their prospects for enhancing human capital in the public sector. In particular, the following aspects were addressed: (a) the current status of AI in relation to human resource management was analyzed; (b) the present and future impact of AI on core areas of human resource management was evaluated; (c) the main challenges posed by AI in matters such as public values, equity, and traditional merit system principles are identified; and (d) the article concluded by highlighting research needs that underscore the growing role and influence of AI applications in the workplace. These applications promise increased efficiency, savings, and effectiveness in public administration and aim to better adapt to the constantly changing current job landscape. For this reason, it is not surprising that these advanced technologies are present in proposals aimed at enhancing human capital in the government sector.

The approach of [34] aimed to measure the constructive outcome of the proposed architecture, addressing issues such as the perceived risk and trust in citizens’ behavioral intentions when using these cognitive AI communication channels. To assess the practical applicability of this design’s science paradigm, action research was conducted, involving the development of an application as a concrete example. In other words, this article combined design theory and methods with behavioral sciences to create an effective communication model, providing valuable insights into the implementation of AI in the public sector.

Ref. [35] proposed a conceptual framework that connects citizens’ perceptions, trust, and intention to follow government-backed, AI-enabled recommendation system suggestions. However, privacy concerns diminished trust, especially when the system requested confidential information from citizens. Additionally, citizens familiar with technology tended to have more trust in recommendations when a function-based communication strategy was employed.

Governments can establish regulations for social media companies, which, in turn, regulate the spread of disinformation on their platforms. Ref. [36] investigated the impacts of initiatives against disinformation, many of which rely on automated decision-making systems using artificial intelligence (AI) to handle the vast amount of content being shared. These impacts were examined from a broader perspective that addressed both illegal online content and the concern for requesting proactive (automated) actions from online intermediaries to enable legal and policy measures.

Regarding security, [37] addressed various topics related to security, architecture, robotics, detection, policies, and operations in the context of the Internet of Things (IoT). This work included information on the latest advances in IoT research from the U.S. Department of Defense, particularly the “Internet of Battle Things“ project. Additionally, it examined the challenges associated with transitioning defense industrial operations to the IoT and provided policy recommendations for regulating IoT use in government settings within free societies.

The unprecedented speed and scope of technological disruption are transforming our personal and professional lives. Public officials are not adequately prepared to address the challenges posed by this exponential change. Many government structures and procedures that have evolved over centuries could become obsolete in the near future. It is essential to lay the groundwork for governments to reevaluate how they can better serve their citizens [38].

Ref. [39] found that, from business applications to everyday situations, artificial intelligence is having a significant impact on how we assess, analyze, and make decisions. AI systems can quickly and accurately analyze large volumes of data, aiding in making decisions based on accurate and relevant information. AI can identify hidden patterns and trends in complex datasets. AI algorithms can process data in real time, enabling faster and more efficient decision-making compared to traditional approaches. These systems can adapt to individual preferences and needs, making decision-making more personalized and tailored to each user or specific situation. They can handle repetitive and routine tasks, freeing up time and resources for humans to focus on more strategic and high-value tasks. As mentioned earlier, artificial intelligence has the ability to learn from previous experiences and continually improve as more information and data become available. By automating certain aspects of decision-making, AI can help reduce human error and increase consistency in decision-making.

Ref. [40] mentioned that the analysis of artificial intelligence methods as a tool to support government decision-making on specific functions is still insufficient. For example, government budgeting can be considered one of the most crucial internal functions of administration, and it is necessary to understand how artificial intelligence can affect this function. This article explores the potential of artificial intelligence techniques to categorize government budget allocations for various programs and policies. Based on the results of this study, we argue that the use of artificial intelligence techniques in government as a data analysis tool can contribute to more effective decision-making in government.

Governments like the government of Egypt have implemented strategies that accelerate and support socioeconomic progress. For this, [41] discussed a significant information infrastructure and decision support systems in various parts of this country, which in turn led towards radical changes in information systems for citizens. These activities are proposed to enable the government to make informed decisions for their proper implementation.

Ref. [42] demonstrated the economic and environmental problems arising from urbanization and water quality degradation in Southeastern California. They identified and evaluated various specific watershed and site-level solutions, considering their effects on basin flooding, recreational opportunities, water quality, and ecological resources. They even proposed alternatives borne from artificial intelligence with the sole purpose of addressing this challenging environmental situation with limited economic resources.

The aim is to enrich the understanding of the role played by AI in knowledge management. Both the possibilities and inherent constraints of fundamental AI technologies are examined and analyzed concerning their capacity to support the knowledge management process. Additionally, ideas and estimations about future research focused on the development of next-generation decision-support environments are shared. Within this context, [43] sought to enrich the understanding of the role played by AI in knowledge management. Both the possibilities and inherent constraints of fundamental AI technologies were examined and analyzed concerning their capacity to support the knowledge management process. Additionally, ideas and estimations about future research focused on the development of next-generation decision support environments were shared.

The use of Big Data (large volumes of data and how they are used) has modernized societies because significant volumes of data are generated daily, and this is expected to increase significantly in the coming years. Faced with the challenge of turning this vast amount of data into useful information, the importance of using advanced technologies such as machine learning has been recognized. Ref. [44] highlights that machine learning is a technology capable of addressing Big Data classification for statistical purposes and even more complex tasks such as decision-making. This aligns perfectly with the vision of Government 3.0, which explores new opportunities to address contemporary challenges by leveraging innovative data-based technologies for decision-making.

The effectiveness of municipal solid waste (MSW) management services is a key indicator of a city’s sustainability. The sustainable development goals (SDGs) have been established as a global roadmap to promote sustainability in various situations, including normal and disaster situations, according to Goal 11, which aims to create sustainable cities and communities through comprehensive disaster risk management. Local governments play a crucial role in the adoption and implementation of local disaster risk reduction strategies. In this context, [45] present findings that can contribute to the development of an appropriate mitigation plan for waste management systems in flood-prone areas, promoting practices that encourage the sustainable development of cities and communities and the achievement of the SDGs.

Ref. [46] examined the impact of public spending in a capital accumulation model based on an adaptation of optimization models. The cited models allow for the identification of the optimal size of the public sector and the composition of state spending that maximizes the growth rate in the transition to a steady state and the long-term per-capita income level. Different allocations of public resources lead to different growth rates during the transition phase, depending on the variables involved. However, it is important to note that the effects of fiscal policy are essentially temporary. Ultimately, we argue that not taking into account the nonlinear nature of the relationship between public spending and growth could distort the results in empirically conducted studies.

Due to fiscal constraints and the increasing demand for transparency, government officials are reviewing the structure and reporting practices of local administrations. However, these efforts are often incomplete, as they do not consider special government levels that are part of the nation. These governments increase government costs, redirect the allocation of limited resources, conceal debt and spending practices from public scrutiny, and reduce the democratic oversight of elected officials. Ref. [47] highlighted public interest concerns regarding these entities with the aim of informing future research focused on this area for improvement and effective development.

Ref. [48] mentioned that social media is replacing traditional media, and being active on them has become an obligation for many organizations. Besides allowing people to express their ideas freely and independently, even though they can create tension, they serve as an uncensored means of expression, enabling institutions to make strategic decisions based on available data analysis. Moreover, the more complex the data, the more valuable it is for the company. According to available data, artificial intelligence can play a significant role as machines replace the need for humans in decision-making and make decisions accordingly. While the European General Data Protection Regulation is beginning to regulate stored data from European citizens, companies continue to store and control vast amounts of user data.

Ref. [49] focused on the relationship between the documentation of social program goals and their execution within the budgetary context. This article emphasized the importance of considering fiscal policy as an instrument for achieving government program goals to assess the level of compliance with social goals using information derived from the experience of results-based budgeting in countries that have implemented it.

Ref. [50] explained that social networks are replacing traditional media. However, recent research often emphasizes the combination of large-scale datasets and high-performance computing technologies, which is often far from the reality for many government agencies that still heavily rely on their legacy systems and time-consuming methods. To overcome this limitation, in this study, we attempt to bridge the gap between cutting-edge artificial intelligence technologies and current government practice by exploring the potential of artificial intelligence and comparing the performance of linear probability, random forest, and deep learning.

Summary of Key Findings from Included Studies

The VOSviewer-generated co-occurrence network of keywords presented in Figure 2 offers a comprehensive visual representation of the key themes and their interconnections within the field of artificial intelligence in government decision-making. This visualization not only confirms the multidisciplinary nature of research in this area but also highlights the complex relationships between various concepts that are central to the discourse on AI in public administration.

At the heart of the network, ”artificial intelligence” emerges as the dominant node, underscoring its pivotal role in the analyzed literature. This central position is closely orbited by ”machine learning” and ”big data,” indicating the technical foundations upon which AI applications in government are built. The prominence of these nodes suggests that much of the current research focuses on the practical implementation of AI technologies, leveraging machine learning algorithms and large-scale data analysis to enhance governmental decision-making processes.

The network reveals a significant cluster formed by ”government,” ”public administration,” and ”decision-making,” highlighting the specific context in which AI is being studied and applied. This cluster’s strong connections to ”artificial intelligence” demonstrate the growing integration of AI technologies into core governmental functions. Additionally, the presence of ”governance,” ”policy,” and ”public policy” as interconnected nodes reflects the broader implications of AI adoption, suggesting that research is not limited to technical implementation but also encompasses the strategic and policy-level considerations of AI in the public sector.

An important theme that emerges from the visualization is the focus on ethical and governance principles, as evidenced by the cluster containing ”accountability,” ”transparency,” and ”discretion.” The prominence of these terms indicates a significant concern within the research community regarding the responsible implementation of AI in government. This cluster’s connections to both technical and administrative nodes suggest that ethical considerations are being integrated into discussions of AI across various aspects of public administration.

The network also highlights the technical aspects of AI implementation through the close association of “automated decision-making” and “algorithms”. This pairing, along with its connections to other key terms, implies a substantial focus on the mechanisms by which AI systems make or support decisions in governmental contexts. The presence of “challenges” as a notable keyword, linked to various other terms, suggests that the literature critically examines the obstacles and potential drawbacks of AI integration in government processes.

Lastly, the appearance of “digital government” and “e-government” in the network illustrates how AI is being conceptualized within the broader context of digital transformation in the public sector. These terms’ connections to other nodes indicate that AI is not viewed in isolation but as part of a larger technological shift in how governments operate and interact with citizens.

Figure 3 presents a network visualization of international collaboration and the research focus in the field of artificial intelligence in government decision-making, generated using VOSviewer (https://www.vosviewer.com/). This graph offers valuable insights into the global landscape of research in this domain, highlighting key players and evolving trends from 2020 to 2023. The nodes represent countries, with their size indicating the volume of research output, while the connecting lines demonstrate collaborative relationships between nations. The color spectrum, ranging from purple (2020) to yellow (2023), illustrates the temporal evolution of research activities.

The United States emerges as the central hub in this network, represented by the largest node with numerous connections to other countries. This positioning underscores the USA’s significant contribution to the field and its pivotal role in fostering international collaborations. Strong links are visible between the USA and other major research centers such as England, Australia, and Brazil, indicating robust collaborative efforts among these nations. The prominence of these connections suggests a concentration of research expertise and resources in these countries, potentially driving the global research agenda in AI for government decision-making.

Interestingly, the visualization reveals an evolving landscape of the research focus over time. Countries like South Korea and India are represented in lighter shades, suggesting more recent and growing contributions to the field. This trend may indicate an increasing global interest in AI applications for governance, with emerging economies playing an increasingly significant role. In contrast, nodes representing countries like England and Australia appear in darker shades, implying their established and ongoing involvement in this research area over the examined period.

The network also highlights some unexpected patterns. For instance, despite its significant technological advancements, Japan is notably absent from this visualization. This could suggest either a focus on domestic research or a potential gap in international collaboration in this specific field. Similarly, the presence of countries like Mexico and South Africa, albeit with smaller nodes, indicates a broadening geographical spread of the research interest in AI for government decision-making, extending beyond traditional tech hubs.

This visualization not only maps the current state of international collaboration in the field but also hints at future trends. The varying node sizes and connection strengths suggest disparities in research output and collaborative intensity among different countries. As the color spectrum indicates a progression towards more recent years, it is evident that the field is dynamic, with new players emerging and established ones continuing to evolve their research focus. This global perspective is crucial for understanding the diverse approaches and priorities in implementing AI in government decision-making across different national contexts.

Figure 4 presents a bibliometric network visualization of author collaborations and research influences in the field of artificial intelligence in government decision-making, generated using VOSviewer. This network map offers valuable insights into the key contributors and evolving research trends from 2017 to 2024. The nodes represent individual authors, with their size indicating the volume and impact of their publications. The connecting lines demonstrate collaborative relationships between researchers, while the color spectrum, ranging from dark blue (2019) to yellow (2024), illustrates the temporal progression of research activities.

At the center of the network, we observe several prominent nodes representing highly influential authors in the field. Dwivedi (2021) and Zhang (2021b) stand out as major contributors, indicated by their large node sizes and central positions. Their prominence suggests that these researchers have produced significant work that has shaped the discourse on AI in government decision-making. The strong connections between these central nodes and other researchers indicate a high degree of collaboration and knowledge-sharing within the field.

The visualization reveals an interesting temporal evolution of the research focus. Authors such as Coglianese (2017) and Berman (2018), represented by darker-colored nodes, appear to have laid foundational work in the field. In contrast, authors like Lamaysek (2023) and Robles (2023), depicted in lighter shades, represent more recent contributions, potentially bringing new perspectives or addressing emerging challenges in the application of AI in governance.

The network structure also highlights several distinct clusters of researchers, suggesting the existence of specialized subfields or research groups within the broader domain of AI in government decision-making. For instance, the cluster around Busuioc (2021) and De Bruijn (2022) might represent a focus on specific aspects of AI implementation in public administration, while the group including Valle-Cruz (2020) and Liu (2019) could be exploring different dimensions of the topic.

This bibliometric analysis not only maps the current state of research collaboration in the field but also hints at future trends. The varying node sizes and connection strengths suggest disparities in the research output and collaborative intensity among different researchers. The presence of nodes dated to 2024, such as those of Gaozhao (2024) and O’Connor (2024), indicates ongoing and future research directions, highlighting the dynamic and evolving nature of this field.

The visualization provides a comprehensive overview of the research landscape, demonstrating the interdisciplinary nature of AI in government decision-making. It shows how the field has evolved from earlier works focusing on foundational concepts to more recent studies that likely address specific applications, challenges, and ethical considerations of AI in governance. This network analysis is crucial for understanding the intellectual structure of the field, identifying key influencers, and recognizing emerging research fronts in the rapidly advancing domain of AI in public sector decision-making.

4. Discussion

After reviewing the 50 selected articles, it is possible to provide a scientific assessment of artificial intelligence and government decision-making. Subsequently, the questions posed in Section 2 are addressed. (Table 1).

4.1. Answer to Question 1—RQ1

Currently, the use of artificial intelligence (AI) in administrative decision-making in the governmental context is steadily increasing. Despite this upward trend, research that has explicitly explored its relationship with good management is scarce. There is a possibility that the full implementation of automated administrative decision-making may be seen as a form of ”robotic governance,” which could entail risks to good management. Therefore, it is essential not to marginalize the contribution of the human workforce but, rather, to promote their collaboration. Relying too heavily on AI in decision-making could reduce human productivity.

However, the introduction of artificial intelligence into government decision-making has proven to be beneficial in several aspects. It has allowed for the reduction in bureaucratic processes and more efficient access to services, increasing the perception of effectiveness and efficiency among the public. AI systems have empowered governments to process large volumes of data accurately and swiftly, resulting in more informed and equitable decisions and strengthening the perception of impartiality.

However, the implementation of AI in government decision-making also presents challenges. The lack of understanding about how AI algorithms work, as well as the possibility of bias in the data used to train these systems, raises concerns. Consequently, it is imperative for governments to address these challenges and commit to using AI in an ethical and responsible manner, ensuring that the public’s trust does not diminish but, on the contrary, is strengthened.

The integration of artificial intelligence (AI) into governmental decision-making processes presents a significant opportunity to enhance public confidence in the government’s ability to address complex challenges. By leveraging AI technologies to expedite data analysis, identify intricate patterns, and offer recommendations grounded in real-time and accurate information, governments can optimize the efficiency and effectiveness of their decision-making mechanisms. This transformative shift towards data-driven governance not only streamlines administrative processes but also fosters a sense of trust and reliability among the public, as they perceive the government’s capacity to address complex societal issues promptly and effectively.

Furthermore, the utilization of AI in governmental decision-making not only enhances operational efficiency but also promotes a culture of transparency and accountability. AI systems’ ability to process vast amounts of data swiftly and accurately empowers governments to make informed decisions based on evidence and insights derived from complex datasets. This data-driven approach optimizes resource allocation and ensures that public services are managed efficiently, meeting the diverse needs of citizens in a timely manner. Consequently, the integration of AI technologies in governmental decision-making processes can lead to improved service delivery, increased public satisfaction, and ultimately, strengthened confidence in the government’s problem-solving capabilities.

The adoption of artificial intelligence in governmental decision-making represents a paradigm shift towards more effective and responsive governance. By harnessing the power of AI to enhance decision-making processes, governments can not only improve operational efficiency but also strengthen public trust and confidence in their ability to tackle pressing issues. The strategic implementation of AI technologies holds the promise of fostering a more transparent, accountable, and citizen-centric approach to governance, ultimately contributing to a more resilient and adaptive government machinery.

4.2. Answer to Question 2—RQ2

Although artificial intelligence is more commonly applied in the healthcare sector, it is currently being implemented in various relevant sectors, leading to the creation of public policies for environmental preservation due to the evident gradual depletion of natural resources. One of the most notable advantages of these tools is their capacity to analyze and process large volumes of data.

Recently, the European Union has launched a program aimed at promoting the European supercomputing ecosystem to boost scientific research and enhance the quality of life for its citizens. Nine of these projects focused on objectives related to the environment and sustainability. However, despite the evident benefits of artificial intelligence, it is crucial to highlight that this technology is in its early stages of development, raising concerns in the environmental domain. The significant energy consumption and time required to train AI models can have a substantial impact on carbon emissions, exacerbating challenges related to climate change.

Currently, efforts are being made to develop environmentally sustainable artificial intelligence technologies, minimizing energy usage and reducing the carbon footprint. Choosing the appropriate AI model architecture can reduce energy consumption by up to 90%, which is a positive step towards environmental conservation, one of the Sustainable Development Goals (SDGs).

In addition to the environmental domain, artificial intelligence is finding its way into a wide range of public policy areas, demonstrating its versatility and potential to address complex societal challenges. In the realm of social policy, for instance, AI-powered tools are being employed to enhance the delivery of social services, such as identifying individuals in need of assistance, automating benefit applications, and optimizing resource allocation. This is particularly crucial in addressing issues like poverty, inequality, and social exclusion.

Furthermore, artificial intelligence is proving instrumental in the development of more efficient and effective transportation systems, contributing to the creation of smart cities. By analyzing real-time traffic data, AI algorithms can optimize traffic flow, reduce congestion, and enhance public transportation efficiency. This not only reduces travel time but also minimizes environmental impact through reduced emissions.

The application of AI in public policy extends beyond specific sectors, encompassing the development of more effective and efficient governance structures. For example, AI-powered chatbots are being deployed to provide citizens with streamlined access to information and services, enhancing government transparency and accountability. The use of AI in public procurement processes is also gaining traction, with the potential to optimize resource allocation, identify potential fraud, and improve overall efficiency.

It is important to note that, while the potential of AI in public policy is immense, its implementation requires a careful consideration of ethical, legal, and societal implications. The development of appropriate regulatory frameworks and ethical guidelines is essential to ensure the responsible and equitable use of AI in policymaking. Transparency and accountability in the use of AI algorithms are paramount to building public trust and ensuring the technology’s long-term viability in the public sphere.

4.3. Answer to Question 3—RQ3

Artificial intelligence (AI) is currently one of the most widely used mechanisms worldwide. Several countries have implemented national policies and guidelines to address AI, balancing its advantages and challenges in international efforts to regulate cutting-edge technologies. AI, along with its associated advanced technologies, has the potential to provide benefits to society as a whole, including individuals, businesses, and governments. However, this promising landscape also raises ethical, legal, social, and political issues of great relevance.

In response to recent and concerning incidents related to deep learning algorithms, the spread of fake news and AI-based political misinformation, as well as ethical dilemmas associated with autonomous systems and challenges in algorithmic decision-making, both national and international organizations have developed policies, strategies, and action plans to address the risks and benefits of AI. According to the Organization for Economic Cooperation and Development (OECD) AI Observatory, more than 60 countries have created their own national AI policies to oversee and regulate this double-edged technology, and this number continues to grow.

Due to differences in economic, social, and political conditions among countries, these AI policies and initiatives have been developed with the consideration of national regulations and international standards. Most previous research has focused on various perspectives and narratives in various regions of the world, especially in Europe and developed nations.

The European Union (EU) has been at the forefront of AI regulation, adopting a comprehensive approach that prioritizes ethical considerations and human-centric principles. The EU’s “Ethics Guidelines for Trustworthy AI“ provide a framework for developing and deploying AI systems in a responsible manner, emphasizing principles such as human oversight, fairness, transparency, and accountability. The EU’s General Data Protection Regulation (GDPR) also plays a crucial role in safeguarding data privacy and ensuring the ethical use of personal data in AI applications.

In the United States, the focus has been on promoting innovation and fostering a competitive AI landscape. The White House Office of Science and Technology Policy (OSTP) has released guidelines for the development and use of AI, emphasizing the importance of responsible innovation and addressing potential risks. The National Institute of Standards and Technology (NIST) has also developed a framework for AI risk management, outlining key considerations for organizations developing and deploying AI systems.

China, on the other hand, has adopted a more centralized approach to AI regulation, prioritizing national security and economic competitiveness. The Chinese government has issued guidelines for the development and application of AI, emphasizing the need for ethical development and promoting the use of AI in strategic sectors. The focus in China is on fostering a robust domestic AI industry and leveraging AI for national development goals.

While different countries are taking varying approaches to AI regulation, there is a growing recognition of the need for international cooperation to address the global challenges and opportunities presented by AI. International organizations like the OECD, the United Nations, and the World Economic Forum are playing an increasingly important role in facilitating dialogue and collaboration among countries to develop common principles and standards for the ethical and responsible development and use of AI.

As AI continues to evolve rapidly, the need for robust and effective regulatory frameworks will only increase. The challenge lies in striking a balance between promoting innovation and ensuring responsible AI development and deployment. International cooperation, transparency, and continuous dialogue among stakeholders will be crucial for navigating the complex landscape of AI regulation and ensuring its beneficial impact on society.

4.4. Answer to Question 4—RQ4

The implementation of artificial intelligence systems in government decision-making processes presents a complex array of challenges and ethical considerations. One of the primary technical obstacles is the integration of AI systems with existing government infrastructure, which often involves legacy systems and disparate data sources. This integration requires significant resources and expertise, potentially creating a digital divide between well-funded and under-resourced government entities. Moreover, the need for high-quality, unbiased data to train AI models poses another technical challenge, as government datasets may be incomplete or outdated, and they may reflect historical biases.

From a social perspective, the introduction of AI in government decision-making raises concerns about job displacement and the changing nature of public sector work. There is a pressing need for upskilling and reskilling programs to prepare government employees for roles that complement AI systems, rather than competing with them. Additionally, public acceptance and trust in AI-driven decisions remain significant hurdles. Citizens may be skeptical of algorithmic governance, particularly if they perceive it as opaque or detached from human judgment.

The transparency and explainability of AI systems emerge as critical ethical considerations. The “black box“ nature of some advanced AI algorithms can make it difficult to understand and justify decisions, potentially undermining principles of democratic governance and accountability. This lack of transparency can also hinder the ability to detect and correct biases in AI-driven decisions, which may disproportionately affect marginalized communities. Researchers have emphasized the need for interpretable AI models and clear mechanisms for challenging and appealing AI-generated decisions in government contexts.

Accountability presents another significant ethical challenge. Determining responsibility when AI systems contribute to or make decisions that lead to negative outcomes is a complex issue. This becomes particularly problematic in areas such as criminal justice, welfare distribution, or public resource allocation, where decisions can have profound impacts on individuals’ lives. Establishing clear lines of accountability and developing robust oversight mechanisms are essential to ensure that AI systems align with democratic values and public interests.

The potential for AI to amplify existing biases or create new forms of discrimination is a pressing ethical concern. AI systems trained on historical data may perpetuate or exacerbate societal inequalities, leading to unfair or discriminatory outcomes in government services and decision-making. Addressing this issue requires not only technical solutions, such as bias detection and mitigation algorithms, but also diverse and interdisciplinary teams involved in AI development and deployment. Ensuring representation from various societal groups in the design and implementation of AI systems is crucial for creating more equitable and inclusive government services.

5. Conclusions

In recent years, the adoption of artificial intelligence (AI) in government decision-making on a global scale has been steadily increasing. Many governments have expressed interest in investing in AI-based solutions to enhance data processing and make more efficient and effective decisions to address growing societal needs. An analysis of literature from relevant sources over the past five years shows that AI has made significant advancements in various fields, although it also faces significant challenges related to ethical considerations, privacy, security, and regulation. Striking a balance between fostering innovation and protecting the fundamental rights and values of society is crucial for the continued development of AI.

As a result, the integration of artificial intelligence into government decision-making is an evolving process. As technologies advance and policies and practices are refined, it is essential to maintain a balance between driving innovation and safeguarding the fundamental rights and values of society. Collaboration between researchers, policymakers, and society as a whole plays a crucial role in shaping a future where AI will become a valuable tool for government decision-making.

Future research directions in the field of AI in government decision-making should address several key areas to advance our understanding and improve the practical implementation of these technologies. One critical avenue for future work is the development of comprehensive frameworks for ethical AI governance. While current research has identified ethical concerns, there is a need for studies that propose and empirically test governance models that balance innovation with accountability, transparency, and fairness in public-sector AI applications.

Another important direction is the exploration of AI’s role in enhancing participatory democracy and citizen engagement. Future studies should investigate how AI can be leveraged to facilitate more inclusive policy-making processes, improve public consultation mechanisms, and bridge the gap between government decisions and citizen preferences. This could involve research into AI-powered platforms for citizen feedback, sentiment analysis of public opinions, and the use of AI in processing and incorporating diverse public inputs into policy formulation.

The long-term societal impacts of AI-driven government decision-making represent a crucial area for longitudinal studies. Researchers should design and conduct multi-year investigations to assess how the increasing use of AI in governance affects public trust, civic participation, and the overall quality of democratic processes. Such studies could provide valuable insights into the potential unintended consequences of AI adoption and help in developing strategies to mitigate negative outcomes while maximizing benefits.

Additionally, there is a pressing need for comparative studies examining AI implementation across different cultural, political, and economic contexts. While much of the current research focuses on developed nations, understanding how AI can be adapted and implemented in diverse governmental systems, particularly in developing countries, is crucial for global progress in this field. Such studies could provide valuable insights into best practices for AI adoption that are sensitive to local governance structures, resource constraints, and cultural norms.

Future work should also address the challenges of AI implementation in crisis management and disaster response within governmental contexts. Research into how AI can enhance government responsiveness, resource allocation, and decision-making during emergencies and natural disasters could yield significant practical benefits. This could include developing AI models for predicting and mitigating the impacts of crises, as well as optimizing the coordination of multi-agency responses.