The "AI uses too much water and electricity" argument isn't completely wrong, but it's aimed at the wrong target.
-
The "AI uses too much water and electricity" argument isn't completely wrong, but it's aimed at the wrong target.
AI doesn't require water-cooled data centers. Lazy, cost-cutting facility operators do.
The real question we need to be asking: How do we regulate responsible cooling and energy sourcing for data centers?
The nuance-free "ALL AI IS EVIL" approach just hands the people who actually want to cause ecocide an easy win.
-
The "AI uses too much water and electricity" argument isn't completely wrong, but it's aimed at the wrong target.
AI doesn't require water-cooled data centers. Lazy, cost-cutting facility operators do.
The real question we need to be asking: How do we regulate responsible cooling and energy sourcing for data centers?
The nuance-free "ALL AI IS EVIL" approach just hands the people who actually want to cause ecocide an easy win.
@xenophile Legislators are ever outdistanced by innovators, exploitative or benevolent. Sometimes they rush to catch up with the taxation, more slowly with market safety and fairness, but they always lag behind.
Sometime we cannot remedy the root cause. Curing symptoms may be the best we can hope for.
Some AI is already great, at least it’s saved mh butt on occasion. But most of it so far is a poor bang-for-buck, and we should stop blatantly destructive experimentation, right?
-
@xenophile Legislators are ever outdistanced by innovators, exploitative or benevolent. Sometimes they rush to catch up with the taxation, more slowly with market safety and fairness, but they always lag behind.
Sometime we cannot remedy the root cause. Curing symptoms may be the best we can hope for.
Some AI is already great, at least it’s saved mh butt on occasion. But most of it so far is a poor bang-for-buck, and we should stop blatantly destructive experimentation, right?
I disagree that curing symptoms is the best we can hope for. Law-as-code works: you build compliance into the architecture, not bolt it on after.
EU AI Act Article 50 transparency can be enforced at the code level: cryptographic audit trails, automatic AI disclosure, immutable records.
Same philosophy I used when authoring redundant networks (now oracle cloud) architecture; You don't write policy to say "please don't crash" you write policy to dictate the requirements.
-
I disagree that curing symptoms is the best we can hope for. Law-as-code works: you build compliance into the architecture, not bolt it on after.
EU AI Act Article 50 transparency can be enforced at the code level: cryptographic audit trails, automatic AI disclosure, immutable records.
Same philosophy I used when authoring redundant networks (now oracle cloud) architecture; You don't write policy to say "please don't crash" you write policy to dictate the requirements.
@xenophile You know I agree, that’s the sane and robust approach. It bugs me how many opportunists press their way into the cracks of necessity and short term pragmatism, and I can’t ignore them.
-
R relay@relay.infosec.exchange shared this topic
