Artificial intelligence as a complement to traditional anti-corruption approaches: potential and limitations, particularly in public procurement

Within the context of discussions on the potential of digital technologies to expand state capacity, this chapter reviews and analyses the use of artificial intelligence (AI) in anti-corruption initiatives, paying special attention to its interaction with ‘traditional’ anti-corruption approaches. The chapter provides a critical assessment of how anti-corruption technologies can be a useful complement to augment the effectiveness of traditional approaches, especially in the context of public procurement. The chapter first assesses the functionalities of anti-corruption technologies, stressing their potential and limitations. It does so by paying especial attention to technological, practical, and legal constraints on theoretically possible implementations of anti-corruption technologies, to set reasonable expectations. The chapter then maps the anti-corruption tasks and functions where those technologies can be more readily deployed. It then addresses the enablers needed for the deployment of such anti-corruption technologies, with a special focus on data and institutional capabilities, as key elements of a maturity model for AI adoption in the public sector ultimately reflective of state digital capability. The chapter then focuses on the development of early warning systems (red flags) to identify and prevent corruption in procurement, focusing on a critical evaluation of their use in Paraguay. The critical evaluation identifies risks and opportunities for the improvement of AI-based early warning systems, as well as broader criteria for the development of algorithms in a manner that adds value to traditional anti-corruption efforts. The chapter concludes by extracting lessons learned and policy recommendations on the use of big data and AI for anti-corruption purposes, with a specific focus on key decisions and aspects to be considered in the development of AI-based early warning systems, if they are to augment state capacity to tackle corruption. The chapter is prefaced by a policy-focused executive summary highlighting and signposting the main takeaway messages, to facilitate focused reading.