AI could consume up 3% of world’s electricity the UN warns

Here is a summary of the United Nations report on the environmental costs of AI, with four main key points outlined:

Projected Energy Demand and Emissions: The report estimates that AI’s energy use could double by 2030, consuming 3% of the world’s electricity and generating emissions equivalent to those of the UK. AI will also excessively use water for cooling, surpassing global annual drinking water needs.
Jevons Paradox and Increased Demand: Despite anticipated efficiency improvements, the “Jevons paradox” suggests that efficiency gains will not lead to reductions in overall resource consumption, but instead promote expanded use, resulting in even higher environmental impact.
Calls for Responsible AI Practices: To address these concerns, the report outlines a roadmap advocating for responsible AI practices based on transparency, efficiency, equity, lifecycle responsibility, global cooperation, and sustainable use. It underscores full value-chain governance from mineral sourcing to recycling.
Global Digital Divide and Inequity: The report identifies an infrastructural digital divide where only 32 nations host AI cloud infrastructure, predominantly situated in US and China, warning about the environmental unfair burden shouldered by nations that consume AI but do not control its production.

This summary highlights key aspects of the report, drawing from its principal concerns and recommendations.