Page 57 - ISC PROCEEDINGS 21.4

Third, evidence-based governance. AI supports risk quantification, enabling capital
allocation decisions and plan adjustments to be based on data-driven analysis rather than
subjective judgment.
Together, these three pillars form an integrated theoretical foundation for
advancing Smart PFM, aligning digital technologies with institutional reform to enhance
efficiency, transparency, and sustainability in public investment governance.
2.2. Literature review
The rapid digital transformation of the public sector has intensified interest in
leveraging BDA and AI to enhance public financial management. In the domain of PIE,
where large-scale projects, complex procurement systems, and multi-level governance
structures prevail, traditional monitoring mechanisms often face limitations related to
data fragmentation, delayed reporting, and limited predictive capacity. Recent literature
suggests that integrating BDA and AI into public investment oversight systems can
significantly improve transparency, accountability, and efficiency, thereby fostering smart
public financial governance.
BDA refers to advanced techniques for processing and analyzing high-volume, high-
velocity, and high-variety datasets to generate actionable insights. Against this
background public finance, BDA enables real-time tracking of ST execution, anomaly
detection in disbursement patterns, and cross-referencing of procurement, treasury, and
project performance databases. Studies highlight that data-driven monitoring systems
reduce information asymmetry and enhance evidence-based decision-making, particularly
in capital expenditure management where risks of cost overruns and delays are
substantial.
Artificial Intelligence, encompassing ML, natural language processing, and predictive
modeling, further strengthens supervisory capacity. AI-based models can identify irregular
spending behaviors, forecast project completion risks, and classify procurement contracts
according to fraud probability. Empirical research demonstrates that supervised learning
algorithms improve the detection of abnormal transactions compared to rule-based
auditing systems. Meanwhile, unsupervised learning methods are increasingly used to
uncover hidden patterns in complex financial networks, supporting proactive risk
management.
The convergence of BDA and AI aligns with the broader paradigm of digital
government and smart governance. Smart public financial governance emphasizes
interoperability, automation, and data integration across treasury, ST, and audit systems.
Literature on digital PFM underscores that technological integration enhances fiscal
discipline by enabling continuous monitoring rather than ex-post inspection. In this
framework, AI-driven dashboards and predictive analytics tools serve not only as control
instruments but also as strategic planning aids, informing resource allocation and
investment prioritization.
However, scholars also identify institutional and ethical challenges. Data quality,
interoperability standards, cybersecurity risks, and algorithmic bias remain critical
concerns. Without robust governance frameworks, AI systems may produce opaque or
discriminatory outcomes, undermining public trust. Therefore, successful implementation
requires complementary reforms in legal frameworks, digital infrastructure, and human
capital development. Capacity building in data science within treasury and audit
institutions is frequently cited as a prerequisite for sustainable adoption.

56