L’objectif de cette étude et d’apporter un éclairage scientifique par une évaluation quantifiée des impacts environnementaux du numérique, afin que chacun∙e d’entre nous, citoyen, entreprise, dirigeant politique, puisse prendre la mesure des impacts du numérique et prendre nos responsabilités pour réduire ces impacts.
The rapid growth of artificial intelligence (AI), particularly Large Language Models (LLMs), has raised concerns regarding its global environmental impact that extends beyond greenhouse gas emissions to include consideration of hardware fabrication and end-of-life processes. The opacity from major providers hinders companies’ abilities to evaluate their AI-related environmental impacts and achieve net-zero targets. In this paper, we propose a methodology to estimate the environmental impact of a company’s AI portfolio, providing actionable insights without necessitating extensive AI and Life-Cycle Assessment (LCA) expertise. Results confirm that large generative AI models consume up to 4600x more energy than traditional models. Our modelling approach, which accounts for increased AI usage, hardware computing efficiency, and changes in electricity mix in line with IPCC scenarios, forecasts AI electricity use up to 2030. Under a high adoption scenario, driven by widespread Generative AI and agents adoption associated to increasingly complex models and frameworks, AI electricity use is projected to rise by a factor of 24.4. Mitigating the environmental impact of Generative AI by 2030 requires coordinated efforts across the AI value chain. Isolated measures in hardware efficiency, model efficiency, or grid improvements alone are insufficient. We advocate for standardized environmental assessment frameworks, greater transparency from the all actors of the value chain and the introduction of a “Return on Environment” metric to align AI development with net-zero goals.
Cette étude vise à mettre à jour les données de l’étude menée avec l’Arcep en 2020 sur l’évaluation de l’impact environnemental du numérique en France, aujourd’hui et demain. En effet, n’avait été pris en compte dans les hypothèses de l’étude de 2020, que les data centers situés sur le territoire français. Or une partie importante des usages en France sont hébergés à l’étranger (environ 53 %) ce qui représente des impacts très loin d’être négligeables. Par ailleurs, entre 2020 et 2022, le mix entre les télévisions OLED et LCD-LED a varié au profit des télévisions OLED plus grandes et plus impactantes ainsi que les usages notamment due à l’arrivée massive de l’IA.
La Métropole de Lyon a été sélectionnée pour devenir pilote locale de la feuille de route “France Numérique Ensemble” (FNE), une initiative nationale visant à renforcer l’inclusion numérique sur tous les territoires. Cette feuille de route, élaborée de manière collaborative, fixe des objectifs ambitieux pour garantir l’accompagnement numérique de tous les citoyens, le développement des compétences numériques, et l’accessibilité des outils et services numériques. En tant que territoire pilote, la Métropole de Lyon s’engage à mettre en œuvre des actions concrètes pour réduire la fracture numérique et favoriser l’autonomie numérique de ses habitants.
In recent years,Large Language Models (LLMs) have significantly improved in generating high-quality code, enabling their integration into developers’ Integrated Development Environments (IDEs) as code assistants. These assistants, such as GitHub Copilot, deliver real-time code suggestions and can greatly enhance developers’ productivity. However, the environmental impact of these tools, in particular their energy consumption, remains a key concern. This paper investigates the energy consumption of LLM-based code assistants by simulating developer interactions with GitHub Copilot and analyzing various configuration factors. We collected a dataset of development traces from 20 developers and conducted extensive software project development simulations to measure energy usage under different scenarios. Our findings reveal that the energy consumption and performance of code assistants are influenced by various factors, such as the number of concurrent developers, model size, quantization methods, and the use of streaming. Notably, a substantial portion of generation requests made by GitHub Copilot is either canceled or rejected by developers, indicating a potential area for reducing wasted computations. Based on these findings, we share actionable insights into optimizing configurations for different use cases, demonstrating that careful adjustments can lead to significant energy savings.
La Métropole Aix-Marseille-Provence présente sa feuille de route pour un numérique responsable et vertueux, visant à concilier la transition numérique avec les enjeux écologiques et sociaux du territoire. Ce document stratégique définit les objectifs et actions à mettre en œuvre pour développer un numérique plus sobre, accessible et au service de la transition écologique et sociale du territoire métropolitain.
This innovative research demonstrates how simple addition operations can be used to create more energy-efficient language models without sacrificing performance. The authors propose a novel architecture that significantly reduces computational complexity and energy consumption while maintaining model capabilities. The study provides empirical evidence showing substantial energy savings compared to traditional transformer architectures.
This comprehensive study analyzes the environmental impact of data centers specifically used for AI training and inference. The research provides detailed measurements of energy consumption and carbon emissions from major AI computing facilities. The authors present innovative solutions for reducing the environmental footprint of AI infrastructure, including advanced cooling systems, renewable energy integration, and workload optimization strategies. The paper also introduces new metrics for measuring and comparing the environmental efficiency of different AI computing architectures and deployment strategies.
This comprehensive survey examines various approaches to make the training of large language models more efficient and environmentally sustainable. The research analyzes different techniques including model compression, efficient attention mechanisms, and hardware-aware training strategies that can significantly reduce the computational and energy costs. The authors provide a systematic comparison of different efficiency methods and their impact on model performance, training time, and energy consumption.
This paper provides a comprehensive analysis of the environmental impact of AI systems throughout their lifecycle, from development to deployment and maintenance. The authors examine various strategies for reducing the carbon footprint of AI, including efficient model architectures, green computing practices, and renewable energy usage. The research also presents concrete recommendations for developing and deploying AI systems in an environmentally responsible manner.