AI-Driven Public Administration Could Reinforce Transparency and Accountability in Papua New Guinea

This work is licensed under a Creative Commons Attribution 4.0 International License. The license permits unrestricted use, distribution, and reproduction in any medium, on the condition that users give exact credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if they made any changes. AI-Driven Public Administration Could Reinforce Transparency and Accountability in Papua New Guinea Victor Jima  School of Economics, Wuhan University of Technology, Wuhan, China Alfred Jima Department of Mechanical Engineering, South Dakota State University, South Dakota, United States Abstract Public administration in Papua New Guinea (PNG) faces ongoing challenges, including inefficiencies, accountability issues, and limited transparency. This study explores the potential of artificial intelligence (AI) as a catalyst for governance reform, improving administrative performance when integrated within supportive institutional environments. Using a mixed-method approach, the research combines qualitative insights with quantitative modeling, developing the AI-Governance Transparency–Accountability Model (AIG-TAM). This framework investigates how ethical governance (E), institutional readiness (R), and AI adoption (A) influence transparency (T), accountability (C), and citizen trust (U). Findings from scenario simulations and comparisons with Estonia, Rwanda, and Kenya suggest that technological investments can enhance governance only if paired with strong ethical standards and administrative capacity. For PNG, these insights advocate for an ethics-driven, capacity-building approach focused on participatory oversight, algorithmic accountability, and human-centered design. This research contributes to the discourse on digital governance in small island developing states, demonstrating AI's role in promoting transparency and accountability. Keywords: Artificial Intelligence, Public Administration Reform, Transparency and Accountability, Digital Governance, Policy Implementation. JEL Classification codes: H83, O38, O33, D73, D78, O57, O10, K40, Z18. Suggested citation: Jima, V., & Jima, A. (2025). AI-Driven Public Administration Could Reinforce Transparency and Accountability in Papua New Guinea. European Journal of Management, Economics and Business, 2(6), 194-208. DOI: 10.59324/ejmeb.2025.2(6).14 Introduction Background and Context Papua New Guinea (PNG) is a nation of immense natural and cultural diversity, yet it continues to confront profound governance challenges that impede its developmental progress. Despite repeated reform efforts, the public sector remains hampered by inefficiency, weak accountability mechanisms, and limited transparency (Walton, 2016; Walton & Hushang, 2020). Geographic fragmentation, sociocultural heterogeneity, and institutional legacies of patronage further EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 195 complicate governance, making equitable and effective service delivery an enduring challenge (Tiki, Luke, & Mack, 2021). These systemic shortcomings have perpetuated a cycle of administrative underperformance and eroded public confidence. Corruption remains pervasive, diverting essential resources from critical sectors such as health, education, and infrastructure. Cumbersome bureaucratic procedures delay service delivery, while a lack of accessible information obscures governmental operations from public view. Consequently, many citizens remain disengaged from and distrustful of state institutions (Duncan & Banga, 2018). Against this backdrop, the global proliferation of artificial intelligence (AI) presents new possibilities for public sector innovation (Figure 1). Governments in countries such as Estonia, Rwanda, and Singapore have leveraged AI to enhance decision-making, automate oversight, and improve public service responsiveness (Plantinga, 2024). Through tools like machine learning, natural language processing, and intelligent data analytics, AI has demonstrated its potential to strengthen accountability and operational transparency. For PNG, these technologies offer not merely incremental improvement, but the prospect of foundational reform enabling a shift from reactive, opaque administration to proactive, data-informed, and open governance (Patra & Singh, 2023). Figure 1. Global Examples of AI Integration in Public Administration Problem Statement PNG’s public administration is characterized by three mutually reinforcing deficits: systemic corruption, bureaucratic inefficiency, and inadequate transparency (Adam & Fazekas, 2018). Highprofile cases of misappropriation particularly in procurement and constituency development funds highlight significant weaknesses in oversight and enforcement (Nidhal, 2024). Manual and siloed record-keeping systems impede the monitoring of public expenditures and the detection of fraud. Moreover, accountability institutions such as the Auditor-General’s Office and the Ombudsman Commission operate under severe constraints, including limited data access and analytical capacity. These conditions have undermined progress toward national development goals and international commitments, including SDG 16, which promotes effective, accountable, and inclusive institutions. While past initiatives such as decentralization and digital government programs have sought to strengthen governance, their impacts have been limited (Ugyel, Sause, & Gorea, 2021). The continued reliance on non-integrated, non-automated systems leaves considerable room for procedural manipulation, administrative delays, and discretionary abuse. EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 196 In this context, AI-enabled governance represents a promising avenue for institutional renewal. By automating high-discretion processes in procurement, financial management, and project monitoring, AI can mitigate opportunities for corruption and enhance accountability (Table 1). However, the integration of AI into PNG’s governance landscape is fraught with challenges, including infrastructural gaps, data inadequacies, and ethical risks such as privacy infringement and algorithmic bias (Cheong et al., 2025). Table 1. AI-Enabled Responses to Core Governance Challenges in PNG’s Public Administration Governance Challenge Impact on Public Administration Potential AI-Driven Solution Expected Governance Outcome Corruption in procurement Misuse of public funds and inflated contracts Predictive analytics for anomaly detection in procurement data Reduced graft and improved financial accountability Bureaucratic inefficiency Delays in service delivery and high administrative costs Process automation (RPA) and workflow optimization Faster service delivery and operational efficiency Data fragmentation Lack of evidence-based decision-making Integrated digital data systems powered by AI Centralized access to reliable administrative data Weak oversight institutions Limited monitoring and auditing capabilities AI-assisted auditing and realtime monitoring dashboards Strengthened institutional accountability Limited citizen participation Low trust in government and limited feedback AI-driven citizen engagement platforms (chatbots, NLP tools) Enhanced transparency and participatory governance Research Objectives and Questions This study aims to critically assess the potential of AI technologies to reinforce transparency and accountability in Papua New Guinea’s public administration. Its specific objectives are to: • Assess PNG's governance and digital readiness to evaluate institutional capacity for AI adoption. • Evaluate AI applications like predictive analytics, automated auditing, and open data to boost transparency. • Analyze ethical, institutional, and infrastructural barriers that may limit AI integration in governance. • Develop and validate the AIG-TAM model to quantify technology, ethics, and institutional readiness interactions. • Propose a policy roadmap for responsible AI implementation, aligning with PNG’s sociocultural and administrative context. By studying this inquiry following are some research questions that come into place: • How can AI-driven systems enhance transparency, accountability, and trust in PNG’s public administration? • What institutional, ethical, and infrastructural factors shape the successful integration of AI in PNG’s governance ecosystem? • In what ways do comparative international experiences (e.g., Estonia, Rwanda, Kenya) inform a context-specific framework for AI-enabled governance in PNG? EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 197 • How do ethical governance and institutional readiness condition the impact of AI adoption on governance outcomes within the AIG-TAM model? Theoretical and Conceptual Framework This study is situated at the intersection of two theoretical traditions: the Good Governance Framework and the Technology Acceptance Model (TAM), as shown in Figure 2. Figure 2. Conceptual Model of AI Integration in Public Administration for Governance Reform in Papua New Guinea Following this introduction, the paper is organized into six section (Figure 3). Section 2 reviews relevant literature on AI, governance, and transparency. Section 3 outlines the study’s conceptual foundations. Section 4 details the research methodology. Section 5 presents findings and discussion, including comparative international examples. Section 6 proposes a strategic roadmap for AI integration in PNG. The final section offers conclusions and policy implications. Figure 3. Flow of Research Design: AI-Driven Governance Framework in Papua New Guinea EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 198 Methodology This research employs a qualitative approach based on literature review, policy analysis, and comparative case studies (Creswell & Plano Clark, 2017). Secondary data from official PNG policy documents, academic studies, and international reports were analyzed thematically to understand institutional readiness and ethical challenges associated with AI-driven governance. Qualitative based Methodology This infographic visually presents the study’s methodological structure for exploring AI’s role in enhancing governance in Papua New Guinea. It outlines eight key components ranging from research philosophy and data collection to analysis framework, ethics, and validity using clear icons, balanced borders, and a blue-gray academic palette. The design emphasizes interpretivist inquiry, multi-source data triangulation, and ethical research alignment with OECD AI principles. This study adopts an interpretivist research philosophy, emphasizing that the meanings and implications of artificial intelligence (AI) in governance are shaped by human perceptions, cultural contexts, and institutional realities (Denzin & Lincoln, 2018). In Papua New Guinea (PNG), governance operates within an intricate web of modern bureaucratic systems and traditional leadership structures. Understanding how AI might improve transparency and accountability therefore requires an approach that values local perspectives and interpretive insight rather than technical measurement alone. Figure 4 illustrates the interpretivist foundation of the study, demonstrating how artificial intelligence (AI) technologies interact with human perceptions and institutional contexts to generate interpretive understanding that enhances transparency and accountability in governance. Figure 4: Conceptual Model Linking Interpretivist Philosophy to AI-Governance Inquiry in Papua New Guinea (PNG) The flow from left to right symbolizes the progression from technological inputs to ethical and administrative outcomes within PNG’s socio-political landscape. The study employs an integrated, multi-method qualitative design consisting of three interrelated components: a) Systematic literature review: to synthesize global and regional insights on AI in governance, transparency, and ethics. b) Critical policy and institutional analysis: to examine PNG’s legal, institutional, and infrastructural readiness for AI integration. c) Comparative case study analysis: to extract adaptable lessons from nations with comparable socio-economic and governance contexts. Together, these components provide a comprehensive and triangulated understanding of AI’s potential in public administration (Yin, 2018). The integration of theoretical insights, policy assessments, and international comparisons ensures that the analysis moves fluidly from conceptual abstraction to contextual application and practical implications. The design acknowledges the EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 199 interdependence between global knowledge and local experience. Lessons from other developing and transitional governance systems, such as Rwanda, Kenya, and Estonia, are examined not as prescriptive models but as adaptable frameworks for PNG’s institutional realities. A systematic, multi-phase search protocol was implemented between January and September 2025 to identify relevant academic literature and policy documents (Flick, 2018). The search strategy encompassed multiple knowledge domains and source types to ensure comprehensive coverage while maintaining scholarly rigor. Primary academic databases provided the foundation for identifying peer-reviewed research, including: • Scopus and Web of Science for high-impact international scholarship • SpringerLink for monograph and conference proceeding coverage • Google Scholar for grey literature and emerging research Complementary institutional repositories supplied crucial policy context and regional perspectives: • World Bank Open Knowledge Repository for development governance frameworks • United Nations Development Program Digital Governance Portal for implementation case studies • Transparency International Reports for corruption perception and accountability metrics • International Telecommunication Union ICT Statistics Database for infrastructure and connectivity data The search protocol employed iterative query refinement, combining controlled vocabulary with natural language terms to capture the conceptual breadth of AI governance while maintaining focus on developing contexts and public sector applications. Quantitative Based Model To move from a theoretical discussion to a testable policy framework, this study proposes a formal mathematical model: the “AI-Governance Transparency-Accountability Model (AIGTAM)”. This model synthesizes the core variables from the integrated theoretical framework Good Governance, Technology Acceptance, and AI Ethics into a dynamic system. Its primary objective is to represent the conditional relationships through which AI adoption (A) influences Transparency (T) and Accountability (C) in Papua New Guinea’s public administration, with institutional readiness (R), ethical governance (E), and citizen trust (U) acting as critical moderating and mediating factors. Figure 5. AI-Governance Transparency–Accountability Model (AIG-TAM) EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 200 The model given in Figure 5 shows how AI integration (A) enhances transparency (T) and accountability (C) in governance, mediated by institutional readiness (R), ethical governance (E), and citizen trust (U). Mathematical relationships depict how ethics and readiness amplify AI’s positive effects, ensuring that technological adoption leads to trustworthy, transparent, and accountable public administration. The model operationalizes the key constructs as continuous variables, scaled from 0 (absent/low) to 1 (optimal/high) (Sterman, 2000). Their definitions are summarized in Table 2. Table 2. Key Variables for the AIG-TAM Model Symbol Variable Type Description 𝑨 AI Integration Index Continuous Composite measure of AI adoption level across government functions. 𝑻 Transparency Level Continuous Degree of public data openness, process visibility, and audit accessibility. 𝑪 Accountability Strength Continuous Effectiveness of oversight, compliance mechanisms, and sanctioning. 𝑹 Institutional Readiness Continuous Composite of infrastructure, digital literacy, and policy alignment. 𝑬 Ethical Governance Continuous Reflects algorithmic fairness, explainability, and human oversight. 𝑼 Citizen Trust Level Continuous Public confidence in government institutions and their operations. The structural relationships among these variables are expressed through the following set of equations: Transparency Equation: 𝑇 = 𝛼0+ 𝛼1𝐴 + 𝛼2𝑅 + 𝛼3𝐸 + 𝛼4(𝐴 × 𝐸) + 𝜖𝑇 (1) This equation posits that transparency is a function of direct AI adoption (𝛼1> 0), which is significantly amplified by institutional readiness (𝛼2> 0), both of which are amplified by ethical governance (𝛼3> 0) (Heeks, 2020). Crucially, the interaction term 𝐴 × 𝐸 (𝛼4> 0) captures the moderating effect of ethical governance, indicating that AI's positive impact on transparency is contingent upon its ethical implementation. Accountability Equation: 𝐶 = 𝛽0+ 𝛽1𝑇 + 𝛽2𝐴 + 𝛽3𝐸 + 𝛽4𝑅 + 𝜖𝐶 (2) Accountability is modeled as being reinforced by transparency (𝛽1> 0), demonstrating the synergistic link between these two principles. AI contributes directly (𝛽2> 0) through tools like automated auditing, while ethical governance (𝛽3> 0) and institutional readiness (𝛽4> 0) provide the necessary foundation for accountable systems (Zuiderwijk & Janssen, 2014). This simulation in figure 20 illustrates how accountability (C) evolves over time as a function of transparency (T), AI adoption (A), ethical governance (E), and institutional readiness (R). The strong, steady rise of C reflects its dependence on these governance dimensions, particularly transparency and AI adoption. The model demonstrates that enhanced ethical governance and EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 201 institutional readiness provide the structural foundation for sustainable accountability in AI-driven public systems. Citizen Trust Dynamics: 𝑈𝑡= 𝛾0+ 𝛾1𝑇𝑡+ 𝛾2𝐶𝑡− 𝛾3(1 − 𝐸𝑡) + 𝛾4𝑅𝑡+ 𝜖𝑈 (3) This dynamic equation models citizen trust as an outcome. Trust increases with contemporaneous levels of transparency (𝛾1> 0) and accountability (𝛾2> 0). The term −𝛾3(1 − 𝐸𝑡) (𝛾3> 0) is critical, representing the erosion of trust caused by ethical deficits, such as algorithmic bias or opacity (Grimmelikhuijsen, 2012). AI Adoption Diffusion (Behavioral Dynamics): 𝑑𝐴𝑡 𝑑𝑡 = 𝛿𝐴𝑡(1 − 𝐴𝑡)[𝜆1𝑃𝑈𝑡+ 𝜆2𝑃𝐸𝑂𝑈𝑡+ 𝜆3𝑅𝑡] (4) Drawing on the Technology Acceptance Model (TAM), this logistic growth function models the rate of AI adoption over time. It is driven by technology’s Perceived Usefulness (𝑃𝑈𝑡) and Perceived Ease of Use (𝑃𝐸𝑂𝑈𝑡) within the public sector, with institutional readiness (𝑅𝑡) as a key enabling factor. The term 𝐴𝑡(1 − 𝐴𝑡) ensures adoption follows a realistic S-curve, slowing as it approaches saturation (Davis, 1989). This translation was executed through a systematic crosswalk between distinct qualitative corpora and the model's core variables. The foundational evidence, drawn from a tripartite structure of academic literature (Corpus A), PNG-specific policy documents (Corpus B), and international comparative case studies (Corpus C), provided the substantive basis for this calibration. For instance: • The Institutional Readiness (R) variable draws its conceptual weight and operational range from diagnostic assessments within PNG's National ICT Policy (2020) and the strategic objectives outlined in its Digital Transformation Policy (2020–2030), which collectively articulate the nation's existing administrative and technological capacities. • The Ethical Governance (E) coefficient was refined through a synthesis of international frameworks, notably the OECD AI Principles (2021) and UNDP guidance on AI Ethics (2023), interpreted through the specific lens of PNG's governance traditions and public service ethics. • The AI Integration (A) variable reflects observed gradients of technological adoption, informed by the staged implementation detailed in PNG's e-Government Roadmap (2021) and tempered by lessons from the digital transformation trajectories of nations like Rwanda and Estonia. • The Citizen Trust (U) variable incorporates the longitudinal narrative of public sentiment, as captured in surveys and reports from Transparency International PNG (2018–2024), which document the perceived linkages between governmental transparency and institutional legitimacy. This meticulous mapping ensures that each parameter within the AIG-TAM is imbued with a traceable empirical justification, transforming abstract themes into grounded, data-informed assumptions. Figure 6 shows the heatmap that visualizes the relationship between qualitative governance themes such as digital infrastructure, ethical oversight, policy coherence, and citizen EJMEB (ISSN 3041-2102) | VOLUME 2 | NUMBER 6 | 2025 202 perception and their corresponding quantitative parameters within the AIG-TAM model. Darker shades indicate stronger qualitative influence on parameters like AI Integration (A), Institutional Readiness (R), Ethical Governance (E), and Citizen Trust (U), showing how contextual insights from policy and ethics inform quantitative model calibration. Figure 6. Qualitative-to-Quantitative Influence Heatmap (AIG-TAM Framework) Figure 7 shows the 3D surface plot that illustrates how variations in Ethical Governance (E) and Institutional Readiness (R) jointly influence the Transparency Index (T) within the AIG-TAM framework. The surface demonstrates that higher levels of ethical governance and institutional readiness produce stronger transparency outcomes, reflecting the model’s calibration of qualitative insights into measurable quantitative relationships. The visualization bridges contextual policy data (qualitative) with simulated governance performance (quantitative). Figure 7. Qualitative-to-Quantitative Calibration Simulation