Explaining Administrative Algorithmic Decisions in Light of the Functions of the EU Administrative Duty to Give Reasons

With public administration becoming increasingly algorithm-infused, it is crucial to ask how much should we understand and communicate about the algorithmic decision-making processes involved therein? To answer this question, this article focuses on the EU law general principle of administrative reason-giving. This principle has been developed for ‘traditional’ human-centred administrative decision-making, yet must now be applied in the algorithmic context. To assess what kind of explanations should administrative bodies give for algorithmic decisions, this article first studies the functions of the reason-giving duty. Second, it analyses EU secondary law—the relevant provisions of the General Data Protection Regulation and the Artificial Intelligence Act—to see what legal requirements already exist for explaining algorithmic output, and whether they meet the various functions of the reason-giving duty. It transpires that not all administrative algorithmic decisions fall within the scope of application of EU secondary legislation concerning algorithmic explanations. Some administrative algorithmic decisions are not subject to any specific explanation obligations. This could leave the rights of decision recipients less protected. Hence, this article advocates for developing an explanation standard applicable to all administrative algorithmic decisions; a standard that would be compatible with the EU law general principle of administrative reason-giving.