The uses of artificial intelligence in the public sector, its variable impact and legal categorisation

Abstract

A broader concept of AI is considered appropriate for automated, algorithmic systems in the public sector. The usual public uses and purposes of AI in the EU and the US are described. The variables that determine the impact, level of risk and legal relevance of public algorithmic systems are studied: relevance of use and specific scope of action, qualification as "high risk", mass use and scale or degree of individualisation of decisions. Special attention is paid to the degree of automation of administrative action. After, on the one hand, the analysis focuses on the difficult legal categorisation of administrative actions of adoption, design and development of public IA systems and their procurement, and also analyses the degree of use of formal regulation and, in particular, formal law. On the other hand, it is analyzed the categorisation and possibility of appeal to the use and effects of public algorithms and and it argues in favour of overcoming certain formalisms.