The article critiques the oversight of AI's tangible environmental impact, emphasizing that AI relies on physical resources like energy, water, and hardware, and calls for integrating planetary stewardship into AI development.
AI Doesn’t Live in the Cloud. It Lives on Earth. We talk about AI as though it exists somewhere ethereal. The cloud. Lovely word. Makes everything sound weightless. But AI is not weightless. Behind every prompt, model and miraculous little answer are physical machines, data centres, electricity, cooling, water, chips, minerals, buildings, cables and land. AI may be artificial. Its resources are not. And that makes me wonder whether we are having the AI debate with one rather important participant missing. Nature. We spend enormous amounts of time asking what AI will do to humanity. Will it replace us? Help us? Outsmart us? Become conscious? Become superintelligent? Take our jobs? Cure disease? All perfectly reasonable questions. But perhaps there is another one. What will AI require from the planet in order to become everything we are asking it to become? Because intelligence, artificial or otherwise, does not exist outside the physical world. Humans need food, water, shelter and energy. AI needs energy, hardware, cooling, infrastructure and the materials from which all of that is made. Different needs. Same planet. And this is where I think our obsession with scale becomes interesting. The prevailing assumption seems to be that bigger is progress. More compute. More data centres. More models. More capability. More everything. But nature has this irritating habit of eventually asking: More until when? There are limits somewhere. There always are. Perhaps the question should not simply be whether we can build increasingly powerful AI systems. Perhaps it should also be: What is the environmental price of the capability we are gaining? Who decides whether that price is reasonable? Who benefits from the capability? Who bears the environmental cost? And are those even the same people? There is another part of this that fascinates me. AI itself does not need a beautiful river. It does not sit beneath a tree. It does not breathe polluted air, watch a forest disappear, feel a heatwave or mourn the extinction of a species. Humans do. Other living things do. So environmental stewardship cannot simply be delegated to the technology on the assumption that greater intelligence will somehow produce greater wisdom. Intelligence and stewardship are not the same thing. Perhaps this is another place where the differences between humans and AI matter. AI may help us model ecosystems, optimise energy use, improve climate science and discover solutions we have not thought of. Wonderful. But humans still decide what we build, how much we build, where we build it and what costs we are prepared to impose on the world around us. Which means Mother Nature probably deserves her own guardrail. Maybe the human and AI relationship is not a relationship between two parties at all. Maybe it is a triangle. Humanity. AI. Planet. And none of the first two gets to flourish for very long by ignoring the third. This is one of the questions that keeps bringing me back to Actualarity my framework for AI, focusing on the now rather than abstract notions of the Singularity. Not what AI might become in theory, but what is actually happening now, what it costs, who carries that cost, and what guardrails reality itself imposes. So before we race towards bigger, faster and supposedly superintelligent systems, perhaps somebody should ask the embarrassingly basic question: Can the planet afford our ambitions? And if the answer eventually becomes no, which one of us is intelligent enough to stop?
The article examines and debunks exaggerated claims about AI's environmental impact, using data from the IEA to show that while AI consumes energy and water, its effects are often overstated and context-dependent.
An article comparing AI's water footprint to animal agriculture, arguing that the latter is much more significant and that a misleading statistic about AI water use has gone viral.
The article explores the growing water consumption by AI data centers, its environmental implications, and potential solutions through technology and strategic siting.
This article discusses how AI can be directed toward ecological restoration and planetary health, highlighting applications like ecosystem mapping and precision reforestation while acknowledging the energy and resource costs of AI itself.
The article argues that AI, while useful, is harmful to individuals, organizations, and society due to environmental, intellectual, and human costs, and calls for resistance until these harms are addressed.