Impact Assessment of the Use of Artificial Intelligence in Public Administration

Abstract

The article discusses the issue of regulating artificial intelligence (AI). By referring to the European Commission’s proposal for regulation on AI and its critique, the article demonstrates practical, political, and axiological difficulties in drafting effective regulation of AI. Without it, however, the potential of AI cannot be fully used, legally and transparently, by public administration. The authors of the article present model rules, elaborated together with their research team in the European Law Institute, which could establish procedural standards for conducting AI impact assessments by public bodies. In the authors’ opinion, the impact assessment, if properly applied, can enable more flexible prevention of or response to significant risks arising from the use of artificial intelligence in public administration.