Roos, C. (2025, January 24). IA não pode ser sustentável se o impacto dos data centers não for reconhecido. Exame. https://exame.com/bussola/ia-nao-pode-ser-sustentavel-se-o-impacto-dos-data-centers-nao-for-reconhecido/
This article examines the environmental and social impacts of data centers as a critical yet often overlooked dimension of artificial intelligence governance. It argues that current narratives of “sustainable AI” obscure the material infrastructures that sustain large-scale computational systems, particularly their intensive consumption of water and energy and their disproportionate effects on vulnerable communities. Drawing on global examples, including cases from the United States, Chile, and China, the article highlights how the expansion of data infrastructures can exacerbate existing environmental inequalities and reproduce patterns of resource extraction. Building on emerging debates in AI ethics, the article engages with the concept of a “third wave” of AI ethics to advocate for a broader framework that integrates environmental justice and social responsibility into the assessment of technological systems. It also critiques corporate sustainability practices, such as the use of renewable energy certificates, which may create misleading impressions of environmental responsibility while masking the actual impact of operations. The article concludes that achieving truly sustainable AI requires stronger regulatory frameworks, greater transparency in resource consumption, and governance approaches that center environmental justice and community impact. Without addressing the material and ecological foundations of AI, claims of sustainability risk reinforcing existing inequalities rather than mitigating them.