AI is Taking Water from the Desert
theatlantic.com
theatlantic.com
"Public data hint at the potential toll of this approach. Researchers at UC Riverside estimated last year, for example, that global AI demand could cause data centers to suck up 1.1 trillion to 1.7 trillion gallons of freshwater by 2027."
Is that a lot or a little? How does it compare to other industries? https://www.motherjones.com/environment/2015/01/almonds-nuts... says CA almonds use 3.5e9 m3 == .9 Tgallons of water. So global AI demand will consume 1-2x the amount of water for CA to create almonds. Alternatively if it takes 872x water per unit of wine (https://muse.union.edu/mth-063-01-f18/2018/09/16/how-much-wa...) and the US drinks 33m hectoliters = .87B gallons of wine == .76 Tgallons of water, then AI will globally consume 2x the US consumption of water for wine.
If I had to choose spending water on AI, almonds, or wine, I would choose AI.
Though, I would love to have a dashboard on ChatGPT telling me that this query consumed X almonds worth of water.
Perhaps we will always find a use for this computing power, as long as governments allow it to be built.
I wonder how much clickbait is taking water from the desert. And as much as I'd _love_ to read about how you rode around in an Uber on a hot day, you're not going to get me to sign up for your website with that.
You could probably see the Palo Verde nuclear plant if you climb up on the roof of that data center...
At least AIs have some objective value.