As generative AI becomes increasingly integrated into our daily lives, its environmental impact is a pressing concern that we cannot afford to overlook. Recent studies reveal that the energy consumption of AI models, particularly during their training phases, can be staggeringup to eight times that of traditional computing workloads. This raises critical questions about the sustainability of such technologies as they continue to proliferate in various sectors.

The environmental footprint of AI extends beyond energy use; it also encompasses significant water consumption and carbon emissions. For instance, training models like GPT-3 has been estimated to consume over 1,200 megawatt hours of electricity and generate substantial carbon dioxide emissions. With a quarter of the global population lacking access to clean water, the water usage of data centers poses an additional layer of complexity to the sustainability debate.