Reasons against AI
Do you need reasons in a discussion with a pro AI idiot?
I posted this at renderosity, but am afraid, it won't last long and will be deleted soon.
We all know, how this page is ruled.
But if you feel stuck and need a reason in a discussion, feel free to use this:
AI's resource demands are substantial and growing rapidly, driven by the energy-intensive nature of training and running large models in data centers.
Electricity: AI servers consume vast amounts of electricity—300 terawatt-hours (TWh) annually by 2028 in the U.S., enough to power over 28 million households. This represents a threefold increase from 2023 levels, with AI-specific electricity use projected to grow 150-fold from 2017 to 2028. A single large AI data center can use as much power as 100,000 homes.
Water: Cooling AI servers is extremely water-intensive. The U.S. could require up to 720 billion gallons of water annually just for cooling by 2028—equivalent to the annual indoor water needs of 18.5 million households. This includes 1.3 to 2.4 gallons of water per kWh of energy used for cooling. Some data centers, like those in Phoenix, use about 385 million gallons per year for cooling alone, not including water used to generate electricity.
Other Resources:
Cooling systems rely on fresh, treated water to prevent bacterial growth and blockages; Google-owned centers discharge only 20% of withdrawn water, with the rest lost to evaporation.
Supply chains for AI hardware (e.g., GPUs, microchips) require 8–10 liters of water per chip for manufacturing.
Fossil fuel-powered electricity amplifies water use and carbon emissions, as thermoelectric power plants consume significant water for cooling.
Land and minerals: Data centers require large land areas and rely on rare earth elements and metals, often mined unsustainably.
Key Insight: While a single ChatGPT query uses only ~0.32 milliliters of water (about 1/15th of a teaspoon), the cumulative impact across billions of queries is immense. The environmental cost is especially high in water-stressed regions like Arizona and the Southwest, where data centers compete with agriculture and communities for limited water supplies.
Conclusion: AI’s resource footprint—especially electricity and water—is not negligible, and current growth trends threaten climate goals, water security, and energy infrastructure.