From Clicks to Carbon: The Environmental Toll of Recommender Systems Article Swipe
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.1145/3640457.3688074
· OA: W4403220878
As global warming soars, the need to assess the environmental impact of\nresearch is becoming increasingly urgent. Despite this, few recommender systems\nresearch papers address their environmental impact. In this study, we estimate\nthe environmental impact of recommender systems research by reproducing typical\nexperimental pipelines. Our analysis spans 79 full papers from the 2013 and\n2023 ACM RecSys conferences, comparing traditional "good old-fashioned AI"\nalgorithms with modern deep learning algorithms. We designed and reproduced\nrepresentative experimental pipelines for both years, measuring energy\nconsumption with a hardware energy meter and converting it to CO2 equivalents.\nOur results show that papers using deep learning algorithms emit approximately\n42 times more CO2 equivalents than papers using traditional methods. On\naverage, a single deep learning-based paper generates 3,297 kilograms of CO2\nequivalents - more than the carbon emissions of one person flying from New York\nCity to Melbourne or the amount of CO2 one tree sequesters over 300 years.\n
