Generative AI is reportedly tripling carbon dioxide emissions from data centers
Generative AI is reportedly tripling carbon dioxide emissions from data centers
Research suggest data centers will emit 2.5 billion tons of greenhouse gas by 2030
cross-posted from: https://lemmy.ml/post/20289663
A report from Morgan Stanley suggests the datacenter industry is on track to emit 2.5 billion tons by 2030, which is three times higher than the predictions if generative AI had not come into play.
The extra demand from GenAI will reportedly lead to a rise in emissions from 200 million tons this year to 600 million tons by 2030, thanks largely to the construction of more data centers to keep up with the demand for cloud services.
Between AI and shitcoin mining, these two "technology branches" already consume more power than all the green power added to the grid combined.
It's why humans will always remain de facto slaves to a few masters. Anything that could potentially be advantageous to all life on Earth? Only if the ones at the top get to profit first. No profit? Enjoy scorching to death on hell-planet for the next forty years!
Between AI and shitcoin mining, these two “technology branches” already consume more power than all the green power added to the grid combined.
And your sources? I only did a cursory search, and according to the IEA data centers are responsible for somewhere in the range of 2-6% of electricity demand. Renewables are currently around 30% globally.
I feel like some people are just emotional reactionaries. They see a certain story, and in their own mind they make the story worse than it is, and treat their feelings as fact.
I have no sources on this, or proof that this guy in particular is doing that.
.........wait, am I doing it right now???
Hmmmmm......
Maybe they got confused about total power usage (maybe) being more than the green power added?
Between AI and shitcoin mining, these two "technology branches" already consume more power than all the green power added to the grid combined.
I think you would be shocked if you learned what some other things in our world cost in CO2.
The energy costs of cryptocurrency mining are easy to calculate because the system is extremely transparent. AI is a little muddier, but we know how much big tech is expanding data centers, and we know how many enterprise GPUs Nvidia sells, so we get a decent estimate.
But these things don't actually do as much damage as compared to other things. Imagine how much energy is used for Gaming PCs and consoles. It's probably up there with Crypto and AI if you consider all running consoles and PCs, plus all the multiplayer infrastructure. But we don't have numbers because this is hard to calculate.
And then there's stuff like personal automobiles, that completely blow these other things out-of-the-water.
I dont think much will remain after this extinction event. Do you know how long it takes niches to refill in an ecosystem? We're going to get to a point where industry collapses and we are reset if we survive at all.
AI consumes power, yes, it's projected to triple its environmental impact, yes, but its environmental impact is much less than most other things. If anything, the AI hate train draws angry peoples' focus off the big polluters that matter.
"Arrghle AI is in everything and modern cars track you, I'll just drive a 30+ year old pickup truck because they don't has no AI tracking nonsense"
Oil and gas companies: money
AI and crypto would be fine if we didn't have to put up with capitalism.
No, because the carbon footprint is ridiculously high.
Between AI and shitcoin mining
The difference is that the better shitcoins have been upgraded to be 99% more energy efficient.
But there's no foreseeable limit or escape from the waste of "AI".
Proof of Stake (PoS) is just dollar bonds without regulations. There's no "difficulty adjustment" to minimize profits, so inequality will just get worse and worse.
I remember when scientists were more focused on making AI models smaller and more efficient, and research on generative models was focused on making GANs as robust as possible with very little compute and data.
Now that big companies and rich investors saw the potential for profit in AI the paradigm has shifted to "throw more compute at the wall until something sticks", so it's not surprising it's affecting carbon emissions.
Besides that it's also annoying that most of the time they keep their AIs behind closed doors, and even in the few cases where the weights are released publicly these models are so big that they aren't usable for the vast majority of people, as sometimes even Kaggle can't handle them.
Everyone thought AI was going to kill us via some Terminator-like Skynet.
Nope.
It’s just going to let us kill ourselves via greed and accelerate destroying the environment.
Which search engines give results without an AI generated response?
Startpage and DuckDuckGo, but you might want to disable summaries in the latter's settings.
Every IT company now: we should increase our server costs by 100x to offer unwanted gimmicks that users don't want and aren't willing to pay
And don't trust
Look, i'm not saying that this isn't a problem. My only question is, is this one of those "global warming is because people don't recycle their soda bottles" things? In other words, How concerned should I be about this vs, taking attention away from the energy, beef, and transportation industry?
Very concerned. It’s currently a race who can speed run us to extinction first.
I don't think this is something to focus on. Tech being 40% of all emissions in the US is suspicious, given that in 2021, all industry was 30.1%, and all transportation was 28.5%. And the total emissions in the US was 6.3 billion tons. https://www.ers.usda.gov/data-products/chart-gallery/gallery/chart-detail/?chartId=108623
I don't have more recent data (if it's in the article, I didn't see it at a skim) but I feel like oil, gas, and agriculture are the bigger long-term targets.
Looks like Techradar misunderstood parts of the source story. The projected emissions over the next 10 years is equal to 40% of all US emissions. The Register
It's a percent against the world's emissions. Be concerned less with this. AI is the current hate train whipping boy, and takes the pressure of public focus off the biggest polluters.
Call me surprised.
It's a nice gimmick and sometimes fun but probably not worth it given the state of the planet already.
This is exactly what using AI feels like:
"The only way to interpret statistics is with a healthy dose of skepticism and a thorough understanding of their context."
While people in this thread jump at the opportunity for this slice of statistics to affirm their confirmation biases, intelligent people will ask what the total carbon dioxide output looks like by comparison.
You're quick to imply that this study is bullshit, yet offer no counter argument except "believing statistics is for losers lul"
So where are your sources to refute the article?
Ita also trivial to come to the same conclusion at a smaller scale.
You can run a LLM at home and see the amount of GPU & power resources it takes to compute the larger models. If I ran that full time, your household bill will most likely be 3x alone.
Can't be logical here. This is a topic that's like discussing immigration with Republicans.
Oh yea, this is happening too.
Lesson: only ask AI if you're still stuck after searching and have no colleague around.
This is the "carbon footprint" fallacy created by big oil. We should vote left and unionize until either the external cost of pollution is internalized with pigouvian taxes, or electricity is rationed by a community-owned organization.
Nobody will notice us shooting ourselves in the foot and expecting corporations to do it too. They don't care if we lead by example unilaterally.
There exists an alternative that uses a lot less power. And also that power is going to get spent no matter what anyway.
and also that power is going to get spent no matter what anyway
Yeah but it leads to higher bills for consumers, and generators can get spun down, and it's keeping fossil fuel plants open, etc.
The alternative I was talking about are called employee brains.
From construction of three data centers or everyday use. Weird wording this whole article.
If I run some AI model on my GPU and power my computer via solar power and some batteries, am I actually contributing significantly to GHG emissions?
Like what is the embodied energy of an AI model?
As usual, pundits and scientists confuse what is and what could be with the truth. For example plastic recycling isn't possible because "right now economics don't make it profitable". Meaning capitalism is killing us, not plastics. I suspect the same is true for AI.
The model has to be trained, refined, etc. You running it off grid isn't the entire process, but I agree with you in a different sense.
If not AI, then there would be some other kind of compute taking up server capacity. It's on the data centers to solve this one, not AI.
Does that take into account that AI models will become more efficient with time?
We never actually used large numbers of monkeys paired with typewritters to produce new literature.
Why? Because it would have wasted all the bananas to produce a bunch of shit.
That is all this level of AI is really equivilant to.
Throwing pudding at a wall, deciding if that toss is closer to the goal than before, changing something, then repeating.
Maybe dont waste the resources until the process is more efficient.
Where are their numbers coming from? The central registry of carbon emissions for genAI data centers? They know shit. They're probably shorting Nvidia and are crying over their losses. Fuck Morgan Stanley
This is posted every week with different numbers. I don't care if data centers use a lot of electricity. They run very efficiently and are a necessity for our modern lives. You have no idea what is being run inside those data centers ai is so much more than chatgpt.
On behalf of all humans suffering from your tech bro arrogance: fuck you and your useless tech. I wish this kind of energy consumption becomes illegal, because it is criminal. Can't wait for the bubble to finally pop.
The article mentions generative AI btw, the kind that powers ShitGPT.