Climate And Resource Awareness is Imperative to Achieving Sustainable AI (and Preventing a Global AI Arms Race)

TL;DR

The paper argues that sustainable AI must balance climate awareness with resource awareness to avoid reinforcing inequalities or compromising environmental integrity. It introduces the base–superstructure lens and the Climate and Resource Aware Machine Learning (CARAML) framework to analyze material conditions and guide cross-level actions. Concrete recommendations span individual researchers, the ML community, industry, governments, and global governance, emphasizing metrics, transparency, collaboration, and governance to mitigate an unregulated AI arms race. The work highlights risks of inequitable access and rebound effects, advocating for Pareto-aware, socially and environmentally conscientious AI development with global coordination and accountability.

Abstract

Sustainability encompasses three key facets: economic, environmental, and social. However, the nascent discourse that is emerging on sustainable artificial intelligence (AI) has predominantly focused on the environmental sustainability of AI, often neglecting the economic and social aspects. Achieving truly sustainable AI necessitates addressing the tension between its climate awareness and its social sustainability, which hinges on equitable access to AI development resources. The concept of resource awareness advocates for broader access to the infrastructure required to develop AI, fostering equity in AI innovation. Yet, this push for improving accessibility often overlooks the environmental costs of expanding such resource usage. In this position paper, we argue that reconciling climate and resource awareness is essential to realizing the full potential of sustainable AI. We use the framework of base-superstructure to analyze how the material conditions are influencing the current AI discourse. We also introduce the Climate and Resource Aware Machine Learning (CARAML) framework to address this conflict and propose actionable recommendations spanning individual, community, industry, government, and global levels to achieve sustainable AI.

Figures (6)

• Figure 1: Compute capacity in different countries based on the Top500 listing of supercomputers using data from November, 2024 top5002024compute. Vast regions in the world, primarily in LMICs have almost no large-scale compute capacity which have grown to become necessary for developing latest AI research.
• Figure 2: (left) Proportion of authors based on the geographic location of their affiliations. The plot shows this based on publications at two ML conferences (ICML and NeurIPS) in 2017 and 2022. The countries are classified based on their income groups as defined by World Bank. We notice that HICs dominate these conferences compared to LMICs. Details on the data preparation is provided in \ref{['app:data']}.
• Figure 3: Contextualizing Sustainable AI using the two axes of climate awareness and resource awareness. The position of this paper is that Sustainable AI should be in the top right quadrant where both climate and resource awareness are high. In case of Edge AI, there is some ambiguity in if it actually is climate aware; so we have marked it with a question mark.
• Figure 4: Proportion of non-renewable and renewable energy sources in different regions of the world. Source:irena2024
• Figure 5: The material basis for developing AI consists of raw-materials, monetary investments, labour, and knowledge commodities. This base forms the foundation of AI as an infrastructure which also shapes the socio-political identity of AI -- manifested through policy, regulation, cultural narratives, media and education -- which can be considered the superstructure. The material base shapes the superstructure, which in turn maintains and shapes the base. Both the base and superstructure are not static; they influence each other. This base-superstructure view, developed as part of historical materialism marx1844philosophic, when applied to AI offers useful insights into how we can influence the combined entity to become more sustainable.