The study, published in Discover Global Society, is notable for where it was conducted. Most empirical work on AI in accounting has focused on private firms in developed economies, where regulatory maturity, institutional capacity, and digital infrastructure differ substantially from those in emerging markets. Palestine and Jordan offer a compelling comparative setting: both are pursuing digital transformation initiatives in their public sectors while their governance frameworks are still evolving. Recent evidence from Jordanian municipalities has shown that digital transformation is increasingly relevant to the efficiency of public financial operations, and regional studies have documented the growing role of AI in accounting decision-making within Jordanian organizations. Yet no prior empirical study had systematically examined how distinct AI capabilities influence government accounting performance in these environments, or whether ethical issues function as the bridge linking technology to outcomes.

The research team, led by Naji Alslaibi of Bethlehem University together with colleagues at Al-Ahliyya Amman University, the University of Valencia, and Tikrit University, disaggregated AI adoption into four specific technological capabilities rather than treating it as a single undifferentiated construct. Big Data analytics allows accountants to instantly analyze massive volumes of financial information, supporting real-time decision-making, fraud pattern identification, and redesigned audit processes. Natural Language Processing addresses the unstructured side of accounting work—contracts, invoices, memos, and regulatory documents—automating document review and even translating natural-language accounting notes into correct entries. Machine Learning brings predictive power: trained models detect errors, forecast financial trends, and flag anomalous patterns that human reviewers might miss. Cloud Computing provides the flexible, secure infrastructure that lets financial data be processed, stored, and accessed from anywhere, reducing operating costs and the need for constant infrastructure upgrades.

Methodologically, the study was rigorous for a survey-based design. The researchers distributed an online questionnaire through official administrative channels in public institutions, receiving 630 responses of which 500 were valid and complete. Constructs were measured with multidimensional multi-item Likert scales and validated using Principal Component Analysis with Varimax rotation, which extracted six clean factors corresponding to the four AI technologies, ethical issues, and task performance. Items loaded at 0.60 or higher on their intended factors, Kaiser-Meyer-Olkin sampling adequacy exceeded 0.70, and Cronbach’s alpha coefficients ranged from 0.79 to 0.90, indicating acceptable to strong reliability. Variance inflation factors between 1.04 and 1.08 ruled out multicollinearity problems, and confirmatory factor analysis showed excellent fit, with an RMSEA of 0.018, CFI of 0.993, TLI of 0.992, and SRMR of 0.030.

The hypothesis testing combined regression analysis with structural equation modeling based path analysis, using 5,000 bootstrap replications to infer indirect effects. All four AI capabilities showed positive, statistically significant associations with task performance. In individual models, Cloud Computing showed the strongest coefficient at 0.492, followed by Big Data at 0.449 and Machine Learning at 0.419, with Natural Language Processing smaller but still significant at 0.312. Machine Learning demonstrated the highest individual explanatory power, accounting for 16.1 percent of the variance in task performance on its own. When all four technologies were entered together, coefficients declined due to shared variance but remained positive and significant, and the combined model explained 35.6 percent of the variance in task performance—a substantial figure for research on workplace performance in complex institutional settings.

The mediation results were the study’s most striking contribution. Ethical issues—conceptualized to include transparency, accountability, integrity, and fairness—significantly mediated the relationship between every one of the four AI technologies and task performance. Among the technologies, Machine Learning exhibited the strongest total association with task performance at 0.401, followed by Cloud Computing at 0.378, Big Data at 0.349, and Natural Language Processing at 0.296. Crucially, in each case the indirect effect through ethical issues was comparable to or greater than the direct effect, meaning that a large share of AI’s performance impact operates through ethical governance mechanisms rather than through raw technological capability alone. Ethics also exerted a strong independent direct effect on task performance, with coefficients ranging from 0.462 to 0.504, confirming that ethical awareness matters for accounting outcomes even apart from its mediating role.

Why would ethics carry so much weight? The authors point to the specific vulnerabilities each technology introduces. Big Data raises privacy and data misuse concerns that can hamper performance unless ethical safeguards are applied. Natural Language Processing can embed linguistic bias and misinterpret data, degrading the quality of financial reporting. Machine Learning, for all its predictive power, suffers from opacity—when the learning code cannot be explained, decisions may drift into unethical territory. Cloud Computing introduces data security risks that make ethical compliance a determining factor in preserving credibility and trust. In public sector accounting, where sensitive financial information, algorithmic opacity, data bias, accountability diffusion, and professional independence are all live concerns, these vulnerabilities are amplified. Emerging markets add further complications, including uneven regulatory maturity and limited technical capacity.

The findings carry direct implications for the United Nations Sustainable Development Goals. AI-enabled accounting systems can strengthen public financial governance by enhancing transparency, reducing corruption risks, and improving decision-making efficiency, which aligns with SDG 16’s emphasis on transparent, accountable, and effective institutions, while the technology infrastructure itself supports SDG 9’s promotion of innovation and resilient digital systems. But the study’s evidence shows these benefits are contingent: without ethical safeguards, AI adoption may undermine trust, fairness, and accountability, contradicting the very development objectives it is meant to serve. The authors frame this as a shift from technological determinism to governance-embedded technological performance—the idea that institutional legitimacy, transparency, and accountability condition whether technology is effective at all.

For managers and policymakers in public institutions, the practical message is that investments in AI must be accompanied by structured ethical governance frameworks. The researchers recommend institutionalizing transparency protocols, algorithmic accountability procedures, and data protection mechanisms alongside technological deployment. Capacity-building should integrate ethical training with technical skill development, and regulatory modernization should move beyond technical standards toward ethics-by-design principles embedded within digital accounting systems, including ethical audit trails and compliance monitoring structures. Balanced investment in both technological infrastructure and governance capacity, the evidence suggests, yields superior performance outcomes than technology spending alone.

The authors acknowledge the study’s limitations, which temper but do not erase its conclusions. The cross-sectional design captures relationships at a single point in time and cannot establish causation. The measures are self-reported perceptions, which may be affected by respondent subjectivity and common method bias. The sample was drawn through non-probability channels, limiting strict population generalization, and the empirical setting is restricted to public sectors in Palestine and Jordan. The hypothesis testing also relied on PCA-derived composite variables in a parsimonious path analysis, which simplifies the structural estimation. Future research, the team suggests, should employ longitudinal designs, incorporate archival or supervisor-rated performance indicators, extend comparisons across countries and sectors, and test moderators such as digital readiness, leadership support, and regulatory capacity. Even with these caveats, the core finding stands out as a warning and an opportunity in equal measure: in the race to digitize government finance, the ethical architecture surrounding the machines may matter as much as the machines themselves.

Subject of Research: The relationship between artificial intelligence technologies and accounting task performance in public sector institutions, mediated by ethical issues, in Palestine and Jordan.

Article Title: Artificial intelligence and accounting task performance in public sector institutions through the mediating role of ethical issues in Palestine and Jordan

Article References: Alslaibi, N., Qawasmeh, R., Samara, H., Hussein, A., & Hussein, W. (2026). Artificial intelligence and accounting task performance in public sector institutions through the mediating role of ethical issues in Palestine and Jordan. Discover Global Society, 4(1), Article 241. https://doi.org/10.1007/s44282-026-00592-3

Image Credits: AI Generated

DOI: 10.1007/s44282-026-00592-3

Keywords: artificial intelligence, accounting, public sector, ethics, machine learning, big data, natural language processing, cloud computing, task performance, Palestine, Jordan, structural equation modeling