Efficiency is Not Enough: A Critical Perspective of Environmentally Sustainable AI
Dustin Wright, Christian Igel, Gabrielle Samuel, Raghavendra Selvan
TL;DR
The paper argues that optimizing ML systems for efficiency alone is insufficient to achieve environmental sustainability due to complex interactions across compute, lifecycle decisions, and platform hardware. It introduces three discrepancies: (1) compute efficiency $\neq$ energy efficiency $\neq$ carbon efficiency, (2) efficiency effects across the model life cycle can be counterintuitive and lead to higher emissions, and (3) platform-level impacts (embodied emissions, water use, e-waste) limit the benefits of efficiency improvements. Through analysis of the EC-NAS benchmark and literature on carbon intensity, the authors demonstrate how reductions in one metric do not reliably translate to lower emissions in practice and highlight rebound effects. They propose systems thinking as a holistic framework to account for interdependencies among people, processes, and infrastructure, and to identify leverage points beyond efficiency to sustainably integrate ML as a technology with broader societal goals.
Abstract
Artificial intelligence (AI) is currently spearheaded by machine learning (ML) methods such as deep learning which have accelerated progress on many tasks thought to be out of reach of AI. These recent ML methods are often compute hungry, energy intensive, and result in significant green house gas emissions, a known driver of anthropogenic climate change. Additionally, the platforms on which ML systems run are associated with environmental impacts that go beyond the energy consumption driven carbon emissions. The primary solution lionized by both industry and the ML community to improve the environmental sustainability of ML is to increase the compute and energy efficiency with which ML systems operate. In this perspective, we argue that it is time to look beyond efficiency in order to make ML more environmentally sustainable. We present three high-level discrepancies between the many variables that influence the efficiency of ML and the environmental sustainability of ML. Firstly, we discuss how compute efficiency does not imply energy efficiency or carbon efficiency. Second, we present the unexpected effects of efficiency on operational emissions throughout the ML model life cycle. And, finally, we explore the broader environmental impacts that are not accounted by efficiency. These discrepancies show as to why efficiency alone is not enough to remedy the adverse environmental impacts of ML. Instead, we argue for systems thinking as the next step towards holistically improving the environmental sustainability of ML.