Submitted 6 months ago by boem@lemmy.world to technology@lemmy.world
https://arstechnica.com/ai/2024/04/power-hungry-ai-is-putting-the-hurt-on-global-electricity-supply/
How much of that AI is actually being used for something worthwhile? Most of the AI I see being sold to the public is pointless bullshit. Nobody needs Microsoft Copilot or Bard or whatever for home/personal use; it’s just a new gimmick that’s called something other than what it is to captivate the imagination. It’s fuckin’ Bonzai Buddy 2.0.
These dumbass “AIs” would be better served as NPCs in video games, and instead their using it to just write the story for 'em. 🤦♂️
Those aren’t being sold to the general public, though. Those have value; the shit being put into our phones and operating systems does not.
Can you find an article that better fits my narrative?
Yeah, i hate how LLMs are getting lumped into actually useful machine learning stuff that has been in development for years without all the hype and idiot CEOs trying to use it to layoff useful employees
Preach, friend. I’m so sick of seeing how everything is getting some useless AI service jammed into it for no reason. At best its just a pointless feature that can be ignored. At worst it’s a data and memory vacuum that mines people’s activity in the name of “product improvements”.
[Very sarcasm] We can just burn more coal or just huck poor people into furnaces or something! Wait no, give them cars! Coal rollin’ cars!!! Why are we even pretending that racing toward extinction isn’t our goal? Embrace doom! Down with humanity! Up with constant fucking around with no regard for costs or consequences!
May I please have some painkillers before being tossed into the furnace, though :-\ Or maybe just like, toss some naproxen in after me I guess?
Could someone give me examples about fields where the AI is already useful?
Because right now it looks like the answer to a question no one asked.
I could see how it could be useful for NOC in videogames or as an assistant, but if it involves wasting so much energy, I don’t see the point.
or as an assistant
Who couldn’t use an assistant?
I find GitHub copilot incredibly helpful — almost indispensable.
That’s where I can’t agree with you. I’d say we need to produce the same amount of energy while trying to have more efficient devices.
That’s also why I’m asking what AI could bring to humanity. If it brings us ways to save energy and have enough food for a growing population, then I’d consider it. But for now, from what I know, it’s just creating a new need when we are already consuming way too much and destroying our planet.
Instead of creating new needs, we should try to live with less, while still enjoying life.
Cars getting bigger and heavier is clearly true, but they are also consuming less (at least for ICE cars) than before. My Seat Leon Cupra is drinking way less fuel than my old Audi S4 Avant while being more spacious inside. I’m also sure that at one point we’re gonna start having smaller and more efficient electric cars because not everyone wants a huge SUV.
Well, Transcendence from 2014 got this right.
Seems like ai companies should take some of that investor cash and build up their own power supply. Maybe they can be the ones to use the small, modular nuclear power stations I’ve heard a little about. Or go bananas on solar and various energy storage solutions.
Maybe we can ask ai how to make more power so there can be more ai. But then, when it tells us, we just mine more crypto. Fuck ai. Crypto won’t murder us.
So either less AI or more efficient AI, right?
Less AI is not an option. But more electricity is another alternative.
I have a dejavu with Blockchain.
As a deep neural network specialist, yes, to both.
This is the best summary I could come up with:
Electricity supply is becoming the latest chokepoint to threaten the growth of artificial intelligence, according to leading tech industry chiefs, as power-hungry data centers add to the strain on grids around the world.
Amazon, Microsoft and Google parent Alphabet are investing billions of dollars in computing infrastructure as they seek to build out their AI capabilities, including in data centers that typically take several years to plan and construct.
But some of the most popular places for building the facilities, such as northern Virginia, are facing capacity constraints which, in turn, are driving a search for suitable sites in growing data center markets globally.
At present, “we probably don’t have enough capacity available” to run all the facilities that will be required globally by 2030, said Sharma, whose unit is working with chipmaker Nvidia to design centers optimized for AI workloads.
In October, the company said in filings to a Virginia regulator that it was experiencing “significant load growth due to data center development” and that growing power demands presented a “challenge.”
In response to the demand, authorities in jurisdictions including Ireland and the Netherlands have sought to limit new data center developments, while Singapore recently lifted a moratorium.
The original article contains 867 words, the summary contains 199 words. Saved 77%. I’m a bot and I’m open source!
Anybody else notice the first graph goes from 2020 to 1996?
I guess that’s meant to be 2025 since the graph is projected? Pretty funky screw up though.
Imgonnatrythis@sh.itjust.works 6 months ago
Is it time yet to darken the sky?
EdibleFriend@lemmy.world 6 months ago
We’re starting with the ocean and working our way up
FaceDeer@fedia.io 6 months ago
That's not going to help improve the electricity supply.
errer@lemmy.world 6 months ago
Sure it will! It’ll allow us to turn humans…into this https://lemmy.world/pictrs/image/2cb0b4c5-eca5-4e49-b3b7-086a7d4d4022.jpeg