Today, we are proud to share the results of the first ever assessment of the full lifecycle environmental impacts of AI models, conducted in partnership with Carbone 4 and ADEME. With this study, we are not only addressing our own impacts but also hope to contribute to a global environmental standard for AI. We aim to set new standards for environmental accountability in AI, advocating for clear, collective rules based on measurable indicators. By sharing these insights, we hope to inspire others to join us in building a more transparent future for AI. Thanks to Hubblo and Resilio I B Corp certified for their peer review. Read more about our approach and findings here: https://lnkd.in/eYJhF-XT
Excellent ! Tout le monde attendait ça depuis longtemps. Ce qu’il reste à faire est de séparer les contributions du training et l’inférence des modèles. Well done Mistral !
Well done! 👏 👏 🦾
Great work sharing this publicly 👏
¡Bien hecho!
Tebrikler!
Would be valuable to break down the “Model training & inference” category into two distinct parts: “Model Training” and “Model Inference.” That way we could get a much clearer picture of « what’s really driving emissions ? » : Is it AI companies’ race for ever-larger models, or is it our everyday use of these models? But in your study they are both are lumped together, which hides important differences :/ (First comment on Linkedin, because it’s one of the few valuable post ever seen) Mistral AI Carbone 4
Hi Mistral AI. Do you have a link to the full LCA study?
🚀 Geweldige update
AI group Program Director - Deputy Group CTO
1wVery pleased to see that Mistral AI is taking this subject seriously. There is no business viable AI without responsibilité. Measurement and Transparency about the AI environmental AI is essential for that. Anna Peronnin : glad to see the progress on this subject since our last discussion in Paris.