Artificial intelligence is often experienced as an immaterial symbolic power: a prompt is entered, and language, images, code, predictions, summaries, or videos appear. Yet AI is not weightless. It depends on a hidden life-ground of electricity, water, land, minerals, labor, communities, ecosystems, and waste sinks. Building on the United Nations University Institute for Water, Environment and Health report Environmental Cost of AI’s Energy Use: Carbon, Water and Land Footprints, this white paper extends the environmental accounting of AI into a life-coherence framework. It argues that AI’s central governance question is not only how large its carbon, water, and land footprints are, but whether the conversion of life-support into symbolic output expands life-capacity or deepens symbolic excess, dependency, and enclosure.
The paper interprets AI as a hidden metabolism linking prompts, models, data centers, electricity, cooling, minerals, labor, e-waste, and ecological sinks. It distinguishes between model training and inference, highlights the escalating footprint of image and video generation, and examines the justice problem of local costs and distant benefits. It then develops the diagnostic framework of AI as Tool, Oracle, Idol, Enclosure, or Commons, before proposing a life-coherent governance framework centered on purpose, proportionality, transparency, sufficiency, lifecycle responsibility, place-based accountability, community consent, public-interest compute, knowledge integrity, and review and repair. A Caribbean and Small Island Developing States application is included to show how fragile grids, water constraints, climate vulnerability, and digital dependency make life-coherent AI governance especially urgent. The paper concludes with a practical Life-Coherent AI Use Protocol for individuals, institutions, governments, communities, and regional commons-building.
