abusesaffiliationarrow-downarrow-leftarrow-rightarrow-upattack-typeburgerchevron-downchevron-leftchevron-rightchevron-upClock iconclosedeletedevelopment-povertydiscriminationdollardownloademailenvironmentexternal-linkfacebookfiltergenderglobegroupshealthC4067174-3DD9-4B9E-AD64-284FDAAE6338@1xinformation-outlineinformationinstagraminvestment-trade-globalisationissueslabourlanguagesShapeCombined Shapeline, chart, up, arrow, graphLinkedInlocationmap-pinminusnewsorganisationotheroverviewpluspreviewArtboard 185profilerefreshIconnewssearchsecurityPathStock downStock steadyStock uptagticktooltiptwitteruniversalityweb
Article

6 Jun 2019

Author:
Karen Hao, MIT Technology Review

Training of AI models contributes to carbon emissions

"Training a single AI model can emit as much carbon as five cars in their lifetimes," 6 June 2019

The artificial-intelligence industry... has an outsize environmental impact... researchers at the University of Massachusetts, Amherst, performed a life cycle assessment for training several common large AI models. They found that the process can emit more than 626,000 pounds of carbon dioxide equivalent—nearly five times the lifetime emissions of the average American car... “While probably many of us have thought of this in an abstract, vague level, the figures really show the magnitude of the problem,” says Carlos Gómez-Rodríguez [computer scientist, University of A Coruña]... The paper specifically examines the model training process for natural-language processing (NLP)... The approach is computationally expensive—and highly energy intensive... computational and environmental costs of training grew proportionally to model size and then exploded when additional tuning steps were used to increase the model’s final accuracy... the process of building and testing a final paper-worthy model... emitted more than 78,000 pounds [of carbon dioxide]...