Artificial Intelligence models consume an enormous amount of water and emit large amounts of carbon in their production, training, operation, and maintenance.
AI's water usage and carbon emissions are both direct and indirect:
Direct usage includes water required to:
Cool data centres
Produce technical components used for AI models
Indirect usage includes water required to, and emissions created by:
Generate electricity that powers data centres
Li, P., Yang, J., Islam, M.A., Ren, S. (2023). Making AI less “thirsty”: Uncovering and addressing the secret water footprint of ai models. ArXiv. Pg.4. http://arxiv.org/abs/2304.03271
It is difficult to keep track of how much of a carbon footprint AI has and how much water is used because of the lack of reporting and transparency from AI developers
The water and energy needed continually grows along with the sophistication, size, efficiency, and use of the AI model
Li, P., Yang, J., Islam, M. A., & Ren, S. (2023). Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models. arXiv.Org. https://doi.org/10.48550/arxiv.2304.03271
Luccioni, A. S., Viguier, S., & Ligozat, A.-L. (2023). Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model. Journal of Machine Learning Research, 24(253).
Yokoyama, A. M., Ferro, M., de Paula, F. B., Vieira, V. G., & Schulze, B. (2023). Investigating hardware and software aspects in the energy consumption of machine learning: A green AI-centric analysis. Concurrency and Computation, 35(24). https://doi.org/10.1002/cpe.7825
Potential Grey Literature/Popular sources: