AI AND TRANSPARENCY IN PUBLIC PROCUREMENT: EVIDENCE FROM ALBANIA AND GLOBAL COMPARISONS

Abstract

Public procurement absorbs a substantial share of public spending—around thirteen percent of GDP across OECD members—and therefore holds immense influence over how effectively governments deliver public goods. Yet procurement is still hampered by hidden information and high monitoring costs. Officials often cannot observe supplier quality or collusion, while firms may struggle to understand evaluation procedures. E-procurement platforms have reduced paperwork but not always corruption. This research asks whether artificial intelligence can narrow those informational gaps and make procurement not only faster but fairer.
To explore that question, the paper constructs the AI Procurement Transparency and Asymmetry Index (AIPTI), grounded in information-asymmetry and transaction-cost theory. The framework groups indicators into four pillars—Disclosure, Competition, Value for Money, and Integrity—that together describe the informational health of a procurement system. The disclosure dimension tracks how completely and promptly governments publish contract data; competition measures bidder diversity and market openness; value for money examines alignment between estimated and awarded prices and the frequency of amendments; integrity reflects how effectively risk-detection algorithms uncover and deter fraud. Equal weighting keeps the index balanced, while validity checks show strong correlation with existing transparency measures such as the CPI and AI Readiness Index.
Applying this framework to Albania, Italy, Greece, Montenegro, Serbia, and North Macedonia reveals a moderate positive relationship between AI readiness and integrity. Italy leads with a higher AIPTI value, followed by Greece, while Albania ranks in the middle—digitally well-equipped yet institutionally constrained. Its high e-government index (0.80) signals solid infrastructure, but its corruption-perception score (42 / 100) shows that technology has not yet translated into deeper trust. The pattern suggests that AI can amplify governance quality only when laws, oversight, and civic engagement keep pace.
International experiences deepen that insight. Korea’s KONEPS system integrates procurement end-to-end and now uses AI for forecasting; Chile’s ChileCompra pairs algorithms with ethical-AI standards; Brazil’s Alice cuts audit time from months to days; and Italy’s ANAC analytics save billions through early detection. Each example confirms that success stems from standardised data and institutional discipline rather than the mere presence of algorithms. Conversely, cases without transparent governance risk turning automation into a new kind of opacity.
For Albania, the lesson is two-fold. The digital foundation exists, and initiatives like Diella—a virtual minister for procurement—signal ambition. What remains is the slower work of legal embedding and skill-building. Strengthening interoperability between agencies, ensuring algorithmic explainability, and opening datasets to civic scrutiny will determine whether AI delivers genuine integrity gains or just procedural novelty. The AIPTI offers a diagnostic path for tracking that progress objectively.
In conclusion, the study finds that artificial intelligence has the potential to convert data abundance into public value. Its real promise lies not in replacing officials but in assisting them—flagging irregularities, improving predictions, and encouraging accountability. When grounded in ethical oversight and human expertise, AI becomes a partner in good governance rather than a black box of decisions. For countries navigating the intersection of digital transformation and institutional reform, this approach provides a measured, evidence-based route toward more transparent, efficient, and trusted public spending.