Think twice before using Generative AI

The environmental impact of using generative AI (genAI) models, particularly large language models (LLMs), which mainstream media inaccurately refers to as “AI,” is not being adequately highlighted. There seems to be no valid reason why mainstream media isn’t emphasizing its environmental consequences, except perhaps that they lack interest (reporting on climate change doesn’t generate views?).

For every generative AI model trained, the enormous amount of electricity required to power servers and the substantial water needed for cooling and preventing overheating continue to be consumed. When these GenAIs are deployed on the cloud, their energy and water consumption continue. Unfortunately, most data centers are not “green,” which means their environmental impact significantly harms the planet.

Articles like “A bottle of water per email: the hidden environmental costs of using AI chatbots” (https://www.washingtonpost.com/technology/2024/09/18/energy-ai-use-electricity-water-data-centers/), “Microsoft’s Hypocrisy on AI” (https://www.theatlantic.com/technology/archive/2024/09/microsoft-ai-oil-contracts/679804/), “Training a single AI model can emit as much carbon as five cars in their lifetimes” (https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/), and “Google’s emissions climb near 50% in five years due to AI energy demand” (https://www.theguardian.com/technology/article/2024/jul/02/google-ai-emissions) demonstrate that these companies prioritize profit over environmental impact. (Let’s not forget that generative AI models are also used to spread misinformation and disinformation, and they consume vast amounts of data without consent.)

I am not against generative AI, but I am not advocating for its widespread use either. I support creating and training these generative AI models (while opposing data collection without consent), but I am concerned about their continued consumption of electricity and water at the expense of our planet and our children’s future. This is why I look for generative AI models that can run locally on my devices. While they do consume electricity (and possibly water for cooling), their impact is negligible compared to running them on the cloud.

Before using any cloud-based generative AI models, consider the consequences. They may impress you with their “intelligence” (which is an illusion) and their captivating voices, but remember that every time you use them, the planet suffers.