Work

Environmental Impact of Artificial Intelligence: Bibliographic Report - Artificial Intelligence and Eco-responsibility Internship

Year: 2023

Type: report

Abstract: The aim of the AI & Eco-design internship was to assess the environmental footprint of AI algorithms and raise awareness among AI practitioners of the impact of their work.This began with an in-depth study of the state of the art on the environmental impact of AI that is presented in this bibliographic report.This study continued for the rest of the internship to help provide a scientific basis for the other tasks we worked on such as (i) measuring and predicting the energy consumed while running inferences on Large Language Models on the Grid'5000 testbed, (ii) exploring solutions that enable users of CEA's clusters to evaluate the environmental impact of their jobs and (iii) proposing directions for CEA to take into account the environmental impact of their AI projects.This report aims to structure our knowledge of the state of the art on the environmental impact of AI and its measurement that we have acquired from our readings.

Source: HAL

Authors: Delort, Etienne, Riou, Laura, Srivastava, Anukriti

Institutions: Commissariat à l'Énergie Atomique et aux Énergies Alternatives, CEA Grenoble, Laboratoire d'Électronique des Technologies de l'Information, CORSE: Compiler Optimization and Run-time SystEms

Language: English

Topic: Green IT and Sustainability

Subfield: Electrical and Electronic Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Climate action

Open Access status: green

OpenAlex ID: W4388677654

Cites: 22

Cited by: 1

Related to: 10