A Taxonomy of AI–Environment Relationships

Abstract Debates on the relationship between AI and the environment—or “sustainable AI”—are often confined to a dualistic framing, pitting AI’s potential as a “savior” for sustainability against its role as a “destroyer” of ecosystems. This article argues that these instrumentalist views are insufficient for capturing the full scope of AI’s relevance to the environment. To provide a more holistic framework, the article introduces a four-part taxonomy that adds two crucial, underexplored roles: AI as a “mediator” and as an “infiltrator.” As a mediator, AI technologies actively shape human perception, knowledge, and moral values concerning the natural world. As an infiltrator, AI moves beyond being a mere tool to become a material and functional component of ecosystems, challenging the boundaries between the artificial and the natural. By analyzing these four roles in conjunction, the article demonstrates the limitations of a purely cost–benefit approach and reframes the central challenge: from how people use AI on the environment to how they might learn to live intelligently with it in the environment.