Assunta Di Vaio, Rohail Hassan, Claude Alavoine
5 min
Over the past two decades, technological breakthroughs associated with Industry 4.0, artificial intelligence (AI), and big data (BD) have fundamentally transformed organizational environments. While private sector firms have widely embraced data-driven decision-making to boost performance and profitability, the public sector has been slower to systematically leverage these capabilities. Data intelligence and analytics (DI&A), when combined with human-artificial intelligence (HAI) interfaces, offer immense potential to optimize government operations, enhance citizen-centric governance, and improve policy outcomes. Despite this potential, academic research on how human-AI collaboration actually functions in public sector decision-making remains fragmented. This study addresses this gap by mapping the existing literature to understand how DI&A and big data contribute to public sector decision-making effectiveness.
The authors conducted a comprehensive bibliometric analysis and systematic literature review using a dataset of 161 English-language articles published between 2017 and 2021. Drawing from major academic repositories including Web of Science, Scopus, and Google Scholar, the study utilized VOSviewer software to map the structure of knowledge production in this field. The investigation evaluated key publication patterns, citation networks, collaborative relationships among authors, and thematic clusters related to DI&A, big data, AI, and human-AI interaction.
To synthesize how public sector agencies adopt and benefit from these technologies, the study integrates three prominent management theories: institutional theory, the resource-based view (RBV), and ambidexterity theory. Institutional theory explains how external stakeholder pressures drive technology adoption, the RBV highlights the importance of internal resource portfolios, and ambidexterity theory addresses how organizations balance exploitation and exploration during decision-making. The bibliometric results indicate that extant literature predominantly concentrates on the standalone technical capabilities of emerging technologies rather than the collaborative dynamics of human-artificial intelligence in public governance. The findings underscore a pressing need for public sector organizations to foster analytical ambidexterity and bridge the gap between technical data analysts and domain-specific decision makers.
Alex: Which raises the obvious question about what a practical intervention actually looks like.
Sam: The paper's synthesis points toward two things. First, public agencies need to develop specific human competencies alongside their technical infrastructure — not just data scientists, but administrators who understand how to interrogate algorithmic outputs, recognize when a model's assumptions don't fit their institutional context, and communicate uncertainty to political stakeholders. Second, organizational culture has to be built to tolerate the ambiguity that data-driven decision-making introduces, rather than demanding false precision from tools that are probabilistic by nature.
Alex: That's a reasonable normative claim. But how much empirical weight does the bibliometric method actually carry?
Sam: That's the right place to push back, and the authors are explicit about the ceiling. Bibliometrics gives you a macroscopic picture of where intellectual attention has been concentrated — it can identify gaps in the literature, but it can't tell you whether closing those gaps would actually improve public sector outcomes. The dataset is also restricted to English-language publications from Scopus and Web of Science, which introduces both language bias and indexing variability. And database records shift over time, so the quantitative snapshot is a snapshot, not a census.
Alex: So the co-occurrence map is evidence of a research gap — not evidence that filling the gap works.
Sam: Precisely. The causal claim — that reorienting research toward human-AI collaboration in governance will improve policy outcomes — is the authors' theoretical inference, not a result the method can directly support. That's why they explicitly call for the next wave of work to move into empirical surveys and longitudinal case studies inside actual public institutions. The bibliometric analysis establishes where the field has been looking; it can't tell you what would happen if it looked elsewhere.
Alex: So taken on its own terms, what does this paper actually contribute?
Sam: A structured, reproducible map of where fourteen years of cross-disciplinary research has concentrated, and a credible case that the concentration is misaligned with the practical problem. The theoretical synthesis — pairing institutional theory with the resource-based view to explain adoption inertia — gives future empirical work a testable framework. That's the genuine contribution: not a solution, but a well-specified problem statement and a theoretical lens for attacking it.
Alex: A useful corrective to a literature that's been talking past the people it's trying to help.
Sam: Thanks for listening to ResearchPod.