Notre contribution pour la création d'un standard environnemental mondial pour l'IA

A comprehensive life cycle analysis was conducted to quantify the environmental impacts of Large Language Models (LLMs), measuring greenhouse gas emissions, water consumption, and resource depletion. The study provides specific metrics for both model training and inference, revealing a strong correlation between a model's size and its environmental footprint. Based on these findings, the company advocates for establishing a global environmental standard for the AI industry to improve transparency and comparability. This would involve standardized reporting on environmental impacts to help users make informed choices and align the sector with climate goals.