The assessment of environmental costs in the public procurement of artificial intelligence systems

Abstract

Although public administrations increasingly experiment with, deploy, and buy artificial intelligence systems, evidence shows that environmental costs are not adequately assessed in the procurement of those systems. The article demonstrates the inadequacy of the normative framework which, in the Italian legal system, currently applies to the public procurement of AI. Such framework, in fact, has not been amended in order to take into adequate account the specificities of AI. What is more relevantly lacking, especially, is the determination of minimal environmental criteria in the field of AI. The article insists on the need to invest in the competences of the human resources employed by public administrations, both in terms of training and in terms of recruitment, in order to pursue greater awareness and sustainability in the public procurement of AI systems.