The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink

2022/06/28 by David S. Patterson, David Patterson, Joseph Gonzalez +13 · 15 citations
Computer Science · #Explainable Artificial Intelligence (XAI)

paper · doi:10.1109/mc.2022.3148714

Abstract

Machine learning (ML) workloads have rapidly grown, raising concerns about their carbon footprint. We show four best practices to reduce ML training energy and carbon dioxide emissions. If the whole ML field adopts best practices, we predict that by 2030, total carbon emissions from training will decline.

Citations

• On the Dangers of Stochastic Parrots
• Green AI

Cited by

• The carbon emissions of writing and illustrating are lower for AI than for humans
• Sociotechnical Implications of Generative Artificial Intelligence for Information Access
• Recent Advances in Natural Language Processing via Large Pre-trained Language Models: A Survey
• The growing energy footprint of artificial intelligence
• Will AI accelerate or delay the race to net-zero emissions?
• To better understand AI’s growing energy use, analysts need a data revolution
• Application of Large Language Models to Software Engineering Tasks: Opportunities, Risks, and Implications
• Hopf physical reservoir computer for reconfigurable sound recognition. [europepmc]
• Material to system-level benchmarking of CMOS-integrated RRAM with ultra-fast switching for low power on-chip learning. [europepmc]
• Machine Learning-Guided Protein Engineering. [europepmc]
• The carbon emissions of writing and illustrating are lower for AI than for humans. [europepmc]
• High-performance deep spiking neural networks with 0.3 spikes per neuron. [europepmc]
• Democratizing cost-effective, agentic artificial intelligence to multilingual medical summarization through knowledge distillation. [europepmc]
• Greenhouse gas emissions due to long-term data storage of CT with reformats and strategies for mitigation. [europepmc]
• Impact of decoding strategies on GPU energy usage in large language model text generation. [europepmc]

Related

• The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink
• Quantifying the Carbon Emissions of Machine Learning
• The Environmental Impacts of Machine Learning Training Keep Rising Evidencing Rebound Effect
• Counting Carbon: A Survey of Factors Influencing the Emissions of Machine Learning
• Carbontracker: Tracking and Predicting the Carbon Footprint of Training Deep Learning Models