Green AI: A systematic review and meta-analysis of its definitions, lifecycle models, hardware and measurement attempts
Marcel Rojahn, Marcus Grum
TL;DR
This paper defines Green AI as a lifecycle-explicit, multi-impact approach distinct from Sustainable AI, and develops a five-phase lifecycle mapped to Life Cycle Assessment (LCA) stages, explicitly incorporating energy, carbon, water, and embodied-material impacts. It then introduces a governance framework based on Plan-Do-Check-Act (PDCA) with Phase Completion Criteria (PCC) and Performance-Environmental Thresholds (PET) to operationalize decisions across provisioning, development, deployment, maintenance, and end-of-life. A unified hardware-software architecture and calibrated measurement framework are proposed, combining direct metering (PMCs, rack PDUs) with calibrated estimators to enable reproducible, cross-provider comparisons across energy, carbon, water, and embodied footprints. Analyses of 103 articles reveal a strong emphasis on training and inference while provisioning and end-of-life are underrepresented, highlighting critical gaps in lifecycle coverage and governance. The work offers actionable guidance, benchmarks, and a research agenda to standardize metrics, enable lifecycle-inclusive reporting, and advance closed-loop calibration between estimators and ground-truth measurements for auditable Green AI practice.
Abstract
Across the Artificial Intelligence (AI) lifecycle - from hardware to development, deployment, and reuse - burdens span energy, carbon, water, and embodied impacts. Cloud provider tools improve transparency but remain heterogeneous and often omit water and value chain effects, limiting comparability and reproducibility. Addressing these multi dimensional burdens requires a lifecycle approach linking phase explicit mapping with system levers (hardware, placement, energy mix, cooling, scheduling) and calibrated measurement across facility, system, device, and workload levels. This article (i) establishes a unified, operational definition of Green AI distinct from Sustainable AI; (ii) formalizes a five phase lifecycle mapped to Life Cycle Assessment (LCA) stages, making energy, carbon, water, and embodied impacts first class; (iii) specifies governance via Plan Do Check Act (PDCA) cycles with decision gateways; (iv) systematizes hardware and system level strategies across the edge cloud continuum to reduce embodied burdens; and (v) defines a calibrated measurement framework combining estimator models with direct metering to enable reproducible, provider agnostic comparisons. Combining definition, lifecycle processes, hardware strategies, and calibrated measurement, this article offers actionable, evidence based guidance for researchers, practitioners, and policymakers.