The Price to Pay for Efficiency: A Legal Study of AI Use in Public Decision-Making

Public administration provides ideal testing grounds for AI tools to improve decision-making. AI can streamline repetitive tasks and provide novel insights into complex problems. However, it also risks creating new issues, sometimes exacerbating existing ones for which citizens must pay the price. The risks linked to AI use can be magnified in the administration, where decision-making is held to a high standard due to public interest. Opaque and erroneous decisions as well as increased State power are the downsides of AI tools that are poorly designed or used inappropriately by public actors. This could undermine the legality, fairness, and trust in public decisions. As AI use by the administration is still experimental, concrete examples are punctual and scattered internationally. Thus, through the study of the experimental use of algorithms or AI models by public actors in France, the U.S.A., Denmark, and India, the present paper contributes to the legal discussion on the use of AI in the public sector. The limits of existing regulatory responses to AI are also discussed. The paper concludes that public administrations’ willingness to use AI for decision-making is proof that their main concern has always been efficiency, despite the recent rise in Europe of strong rival notions like transparency 1 and administrative democracy 2 .