AI Greenwashing? A Comparison of AI Companies’ Environmental Policies and Actual Impact

Abstract

Environmental scientists have been researching and quantifying the negative environmental impact of AI systems for years. Simultaneously, cognizant of the need for good press and consumer concern about environmental impacts of their technologies, companies that specialize in various forms of AI development (CAIDs) self-publish their environmentally friendly plans to “go carbon neutral” or dedication to “sustainable development,” sometimes accompanied by reports about their own energy consumption, water consumption, and pollution generation. The purpose of this paper is to gather those two information points in one place and frame them side by side, the better to highlight discrepancies between what is happening and what AI companies say is happening. This research is comprised of a literature review and a comparative policy analysis study. In presenting both components in one place, I will investigate and clarify the potentially performative, or "Press Release Ethics" (PRE) nature of CAID environmental claims; that is, I will identify greenwashing practices in AI company publications and policies.