| dc.description.abstract | Public procurement is a major part of global GDP, making it very vulnerable to corruption. This has led governments to use Artificial Intelligence (AI) as a key tool for oversight. However, the shift from technical possibilities to real-world effects is often overlooked. This dissertation looks into the factors that influence how well AI-based Anti-Corruption Tools (AI-ACTs) work in public procurement. The effectiveness is assessed through the lens of the ISO 9000 standard, focusing on the extent to which planned anti-corruption outcomes are achieved.
The research uses an Explanatory Multiple-Case Study approach, relying on qualitative methods through Secondary Data Analysis. It examines three specific AI tools: ALICE and MARA in Brazil, and SALER in Spain. These examples create a comparison between a well-established AI environment in Latin America and a developing regulatory system under the EU AI Act. The study tests three hypotheses: (1) that transparency in algorithm design, (2) data quality, and (3) human oversight are key factors that affect AI effectiveness.
The results validate all three hypotheses. The findings show that AI greatly improves detection speed and volume, but its effectiveness is limited by institutional, legal and ethical factors. Transparency (H1) is necessary for legal acceptance, as black box systems encounter considerable judicial pushback. Data quality (H2) serves as a functional limit; biases in training data can reproduce existing administrative issues instead of resolving them. Lastly, human oversight (H3) is crucial for connecting digital alerts to corrective actions, turning algorithmic results into practical integrity measures.
The research concludes that AI is not a quick-fix but rather a socio-technical approach. AI functions as a decision-support tool, with its success relying on strong data governance, legal clarity, human oversight, closing skills gap and respect for human rights. Policy recommendations highlight the importance of explainable AI and breaking down organizational data obstacles to fully realize the benefits of digital oversight. Finally, This study offers a strategic guide for policymakers and future researchers aiming to establish tailored safeguards in the digital battle against public procurement corruption. | en_UK |
