Certainly curious to see sources, but last I checked the industry was an insignificant contributor to electricity use overall. That number is obviously growing, but when I say insignificant, I mean negligible.
Again though, happy to see data that shows otherwise it you have a source to provide.
Energy prices are already rising for people near data centers. Near me, the state is reopening a nuclear plant closed because the maintenance to repair years of use exceeded its value. With DCs moving in, it is now a necessity. As you mentioned, it’s getting worse, but it’s already having an effect.
There’s also the insane water consumption required to cool their stuff. It’s destroying ecosystems everywhere.
From a selfish-ish point, it’s also already increased costs of chips used in RAM and GPUs and has taken stick off the market for consumers.
As far as your open source views, that’s great, but open source projects still get licensed under open source licensing and is at the behest of the creator. Ignoring that is effectively stealing from creators and that’s not okay. It’s one thing to learn from something, and then to cite those sources, and it’s another to take it, regurgitate it and not give credit. AI has been used to impersonate people in music and other media like content creators hurting their income and image with no recourse.
AI is ass at coding. It’s not a good teacher and struggles with any level of complexity. It is ok for troubleshooting, but it has been shown in almost every case that it’s not capable of replacing even junior devs effectively. AWS just had a coming to Jesus moment recently because AI generated code broke critical services and took down services that millions rely on. It’s not security conscious, and there are breaches of personal data left and right.
I’m not saying all uses of AI are the devil. It has it’s place for minor tooling, but the ethical implications mentioned above are just what I care to spend time elaborating on, but there are many more.
To elaborate on the power consumption a little… Current data centers are running at around 100+mw compared to the average household using around 1kw. A single center uses the power equivalent of town 20k-100k in population size. Meta is planning to build a data center that will run at around 2gw, which is a magnitude larger in power running over 1,000,000 GPUs compared to their current 100k+. So one DC is a larger city worth of power. In the grand scheme of the entire power infrastructure in the US, it may not register as something significant, but in the local areas they operate in, it’s often more than the rest of the city combined.
rabiezaater@piefed.social 2 months ago
Certainly curious to see sources, but last I checked the industry was an insignificant contributor to electricity use overall. That number is obviously growing, but when I say insignificant, I mean negligible.
Again though, happy to see data that shows otherwise it you have a source to provide.
BassTurd@lemmy.world 2 months ago
Energy prices are already rising for people near data centers. Near me, the state is reopening a nuclear plant closed because the maintenance to repair years of use exceeded its value. With DCs moving in, it is now a necessity. As you mentioned, it’s getting worse, but it’s already having an effect.
There’s also the insane water consumption required to cool their stuff. It’s destroying ecosystems everywhere.
From a selfish-ish point, it’s also already increased costs of chips used in RAM and GPUs and has taken stick off the market for consumers.
As far as your open source views, that’s great, but open source projects still get licensed under open source licensing and is at the behest of the creator. Ignoring that is effectively stealing from creators and that’s not okay. It’s one thing to learn from something, and then to cite those sources, and it’s another to take it, regurgitate it and not give credit. AI has been used to impersonate people in music and other media like content creators hurting their income and image with no recourse.
AI is ass at coding. It’s not a good teacher and struggles with any level of complexity. It is ok for troubleshooting, but it has been shown in almost every case that it’s not capable of replacing even junior devs effectively. AWS just had a coming to Jesus moment recently because AI generated code broke critical services and took down services that millions rely on. It’s not security conscious, and there are breaches of personal data left and right.
I’m not saying all uses of AI are the devil. It has it’s place for minor tooling, but the ethical implications mentioned above are just what I care to spend time elaborating on, but there are many more.
BassTurd@lemmy.world 2 months ago
To elaborate on the power consumption a little… Current data centers are running at around 100+mw compared to the average household using around 1kw. A single center uses the power equivalent of town 20k-100k in population size. Meta is planning to build a data center that will run at around 2gw, which is a magnitude larger in power running over 1,000,000 GPUs compared to their current 100k+. So one DC is a larger city worth of power. In the grand scheme of the entire power infrastructure in the US, it may not register as something significant, but in the local areas they operate in, it’s often more than the rest of the city combined.