Current debates on AI and sustainability are dichotomised, tending to focus on either the environmental impact of AI systems themselves ("sustainability of AI") or AI's potential for environmental benefit ("AI for sustainability"). This perspective highlights a crucial gap: "AI against sustainability" - the negative environmental consequences stemming from the application of AI technologies. While AI can offer solutions, its use in sectors like fossil fuel extraction or targeted advertising can exacerbate environmental harms, often overlooked in existing discussions. We argue for a systemic understanding of these impacts, distinguishing between AI as an object and its application, and propose a three-pronged approach to tackle "AI against sustainability" by (a) strengthened regulation, (b) proactive self-commitment by industry, and (c) constructive dialogue among stakeholders. Addressing the blind spots of "AI against sustainability" requires moving beyond isolated actions and fostering collaboration across disciplines to ensure truly more sustainable AI.

翻译：暂无翻译