Title: Ignoring AI Sustainability Drives a Global Arms Race

Abstract

Sustainability is generally understood through three primary lenses: economic, environmental, and social. Yet, current conversations regarding sustainable artificial intelligence (AI) tend to concentrate almost exclusively on the environmental dimension, leaving the economic and social pillars largely unaddressed. To realize genuinely sustainable AI, it is crucial to resolve the friction between environmental goals—specifically, reducing the sector’s climate footprint—and social objectives, which prioritize equitable access to the resources required for AI development. However, the drive to broaden access frequently disregards the ecological price of scaling up resource consumption. This position paper contends that balancing climate consciousness with resource awareness is vital for achieving true sustainability in AI; failing to do so exacerbates a global AI arms race. By employing Karl Marx’s base-superstructure model from historical materialism, we examine how underlying material conditions are influencing both the trajectory of AI advancements and the narratives surrounding them. Additionally, we propose the Climate and Resource Aware Machine Learning (CARAML) framework, offering concrete strategies across individual, community, industrial, governmental, and international levels to foster sustainable AI practices.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC