Who Acts, and Who is Watched? Differences Between Top-down and Bottom-up AI Anti-Corruption Tool Implementation Methods

Abstract

Recent data shows stagnation in corruption level, eroding public trust in government institutions. AI offers a valuable opportunity to address corruption. This research aims to explore how AI anti-corruption tools differ based on their implementation methods—top-down or bottom-up—and how these methods affect the relationship between society and government. Key tasks include conducting a bibliometric analysis of AI and corruption, comparing top-down and bottom-up implementation approaches, and reviewing real-world practices in the European Union. The research method includes a literature review of existing global studies, supplemented by data from relevant institutions, while the overview of current AI anti-corruption tools primarily relies on publicly available online information. Findings indicate that while top-down method improve efficiency, it may lead to a "corruption trap." Conversely, bottom-up approach empowers citizens but struggle with open data access. The study highlights the need for further empirical research on integrating open data with AI anti-corruption tools and comprehensive evaluations of existing AI anti-corruption applications.