Artificial Intelligence in Public Organisations to Tackle Public Procurement Corruption: Towards a Theoretical Understanding

AI-based anticorruption tools are greeted with high hopes in the fight against corruption. Thanks to the speed processing of huge volumes of data, they can either ex-ante predict corruption through automated recognition of patterns and ex-post detect corrupt practices uncovering anomalies. Tools of this kind are increasingly adopted by public organizations, especially in public procurement which has always been particularly vulnerable to corrupt practices. However, the “rush to AI” goes hand-in-hand with some skepticism: indeed, AI intrinsically entails severe challenges in terms of technology, regulation, society and ethics. Not only is research on AI-based anticorruption tools extremely topical in both academia and policymaking, but it is also much-needed to better understand them and the possible side effects they might generate. Despite this, current research on the matter is quite scarce and fragmented. This paper has a twofold aim: i) illustrate the current state of the art on top-down AI tools to tackle public procurement corruption in the EU, and ii) set the stage for a theoretical framework to investigate them. Existing evidence on how AI can curb corruption as well as the potential and challenges of these tools will be discussed. By drawing on desk research and document analysis, current applications in place in the EU will also be mapped. An interdisciplinary theoretical framework will then be outlined, starting from the four core constituting elements of these tools - corruption in public procurement, AI in the public sector, ethics, and policy and regulation. Therefore, this paper aims to contribute to corruption and e-government studies by providing a systematized theoretical picture on AI top-down applications to curb public procurement corruption.