Impacts of AI-based anti-corruption audits on risk aversion in decision-making: a case study of the Brazilian ALICE tool

This study examines the influence of AI-based anti-corruption audits on public procurement decision-making, specifically focusing on contract volume fluctuations within the Brazilian federal government. We leverage panel data regression models, utilizing alerts generated by the ALICE platform (Analyzer of Bids, Contracts, and Notices, developed by the Office of the Comptroller General of Brazil) from January 2019 to January 2024. Our findings strongly support that ALICE's AI algorithms can mitigate risk aversion and significantly impact acquisition decisions. Disregarding control variables, alerts are associated with an increase in total procurement by nearly 20% over the period analyzed. While these results suggest a causal relationship between ALICE and changes in decision-making behavior, further research employing qualitative methods, such as in-depth interviews with procurement officials, is necessary to elucidate the underlying mechanisms fully. This study underscores the importance of continued exploration into the complex interplay between AI tools and public sector decision-making. Such investigations are crucial to inform the development and implementation of AI-driven solutions that foster transparency, ethical conduct, efficiency, and accountability in public procurement processes.