It’s so depressing.
Like, I get it’s hard to look inwards. You might discover that you’ve done something that has harmed someone and that you are “bad” (you’re not). If you live in a world that values punishment over repairing harm it’s less risky to keep causing harm than to risk being punished.
You can’t know peace until you are willing to let go of that and truly look at yourself. Acknowledging that speciesm is wrong might mean feeling shame for doing it. It’s astonishing that fear of feeling shame is seen as a worse outcome than what animals go through and the impact that’s having on the environment. Do you want to recon with that choice when the world is burning ?
deleted by creator
Afaiu there’s far more ram, gpu’s and electricity being shat down the drain by wee guys playing computer games than by AI use.
If it wasn’t for the gamers needing to constantly live in some 4k fantasy land the resources for machine learning for actually useful stuff would be legion.
Source and numbers please
Hard to be exact but this estimates around 20 terawatt hours last year spent on AI
In contrast US gamers alone way back in 2019 were shitting 34 terawatt hours down the toilet for gaming.
https://eta.lbl.gov/publications/toward-greener-gaming-estimating
Things may change, but at the moment the gamers are the ones doing more to destroy the planet to live in computer generated hallucinations and help get Gabe more mega-yachts via his propriety crapware store.
The article cites a 2023 figure. The industry was significantly smaller then. The datacenter energy usage changes we’re seeing now are significantly larger than that
Globally, the electricity consumption of data centers rose to 460 terawatt-hours in 2022 […] By 2026, the electricity consumption of data centers is expected to approach 1,050 terawatt-hours
This is also expected to keep going up and up as training energy usage grows ever more. The industry has already delayed fossil fuel phase outs by ramping up demand, and it’s now requiring entirely new capacity to be built
Anthropic estimated that by 2027, training a single frontier AI model will require five gigawatts (GW) of power and projected that the U.S. AI sector alone will require 50 GW of new electric capacity by 2028 to maintain global AI leadership. To put the 50 GW figure in context, it is about twice the peak electricity demand of New York City
[…]
The IEA notes that while a typical data center can consume as much electricity as 100,000 households, the largest next-generation campuses currently under construction will demand 20 times that amount
https://www.brookings.edu/articles/global-energy-demands-within-the-ai-regulatory-landscape/
Before this current AI boom, datacenter usage was almost flat for 20 years and was a small portion of power demand
Wait, what?
Why target gamers? How much electricity have people used watching TV? Charging phones, or cars, or running AC?
How much electric was used by businesses? Or manufacturing? What about street lights?
You can go on and on cherry picking things but AI is far worse for the environment than gamers.
As for ram and GPUs, AI corpos have literally taken over market consumption of them. Manufacturers aren’t making ram or GPUs for the PC market, they’re almost exclusively manufacturing for AI. Which has driven prices up by 3x for anyone not an AI corpo.
Why not target gamers? Why target those that feast on the flesh of suffering?
The article was comparing the environmental cost of AI, and it struck me that people often ignore other forms of high-end computing that consume enormous amounts of resources for little real gain for humanity.
If you want to broaden the discussion to TVs, air conditioning, manufacturing, street lights etc that’s fine, but much of that stuff seems rather important and often focused on efficiency, playing the latest high end games far less so.
The point more that there’s a tendency to single out LLMs while treating enthusiast gaming, or consuming animals, as environmentally invisible. A high-end gaming PC can draw 500–700 W for hours at a time playing games. That’s not trivial, and unlike street lighting or industrial manufacturing it’s largely discretionary. I game and run llm’s on a n100 that draws ~3-20 watts and can manage the overwhelming majority of all games ever made just fine…but why eat beans when you can have steak?
I’m not advocating for the current clusterfuck that is LLM’s development but AI also has legitimate uses in science, medicine, engineering, software development, accessibility, fraud, warfare and research…we need vaccines and to better grok proteins, I don’t think we need 500w gaming rigs and Steam machines any more than we need cheeseburgers and game devs should perhaps consider the actual environment instead of computer generated ones. Efficiency has gone out the window long ago for the pursuit of putting glitter on shite.
This quote from the article is also seems relevant:
“A single query on ChatGPT-3 uses roughly ten times more energy than a typical Google search.”
Even if that’s approximately true, a single LLM interaction often replaces multiple searches and pages of clicking around, Kimi ftw. Looking at energy per query without considering the usefulness of the query isn’t a particularly meaningful comparison. It seemed pretty obvious in the early 2000’s the feeding the beast that is Google was a terrible idea and knowing this people at scale decided to say ‘fuck it’ and feed it until it looks like Mr Creosote and is now vomiting it all back up in technicolor…what on earth did they expect to happen?
I’m just interested in putting AI’s energy use into the wider context rather than treating it as uniquely evil while ignoring other forms of compute power that people happily use every day at scale. The tech is amazing imo but much of it is being trained on wikipedia, reddit, lemmy, 4chan etc as all the good knowledge has been paywalled in ivory towers the past 100yrs by the greedy and thus models are feeding upon the toilet of the internet to avoid being dragged into court.
