The Do Not Significant Harm (DNSH) principle and the environmental impacts of artificial intelligence

Abstract

The European Union believes that digital technology, and in particular Artificial intelligence (AI), can make an essential contribution to the achievement of climate neutrality. However, the environmental footprint of IA seems not adequately addressed. The purpose of this paper is to verify whether the DNSH (Do Not Significant Harm) principle, which constitutes a horizontal conditionality within the Recovery and Resilience Facility, which should be respected also by those measures that involve the use of IA systems, is able to detect and limit the damages that AI could cause to the environment.