Writing a 100-word email using ChatGPT (GPT-4, latest model) consumes 1 x 500ml bottle of water It uses 140Wh of energy, enough for 7 full charges of an iPhone Pro Max
Oh you don’t mean… Oh yeah totally that’s awfulllll like thirsty… Yeah…
Am I going insane? As far as I know cooling with water doesn’t consume the water, it just cycles through the system again. If anyone knows otherwise PLEASE tell me.
Industrial HVAC systems use water towers to cool the hot side of system. The method relies on physics of evaporative cooling to reduce temperatures of the water. The process requires water to be absorbed by atmosphere, to drive the cooling effect. (Lower the humidity, the higher the cooling efficiency is, as the air as greater potential to absorb and hold moisture).
The method is somewhat similar to power station cooling towers. Or even swamp coolers. (An odd example would be, experimental PC water cooling builds with ‘bong coolers’, which are evaporative coolers, built from drainage pipes)
yea i really don’t know when or why they started measuring electricity in water
Maybe it’s a valid measure in the future, albeit 500ml would be enough to power New York for a day (the state) by means of fusion.
perplexity.ai says that one chat GPT query consumes half a liter of water O_O
im imagining a rack of servers just shooting out a fire hose of water directly into the garbage 24 hours a day
These article titles are so crazy. Who thinks of this stuff?
ChatGPT.
140Wh seems off.
It’s possible to run an LLM on a moderately-powered gaming PC (even a Steam Deck).
Those consume power in the range of a few hundred watts and they can generate replies in a seconds, or maybe a minute or so. Power use throttles down when not actually working.
That means a home pc could generate dozens of email-sized texts an hour using a few hundred watt-hours.
I think that the article is missing some factor, such as how many parallel users the racks they’re discussing can support.
You are conveniently ignoring model size here…
Which is a primary impact on power consumption.
And any other processing and augmentation being performed. System prompts and other things that are bloating the token size …etc never mind the fact that you’re getting a response almost immediately for something that an at home GPU cluster (not casual PC) would struggle with for many minutes, this isn’t always a linear scale for power consumption.
You are also ignoring the realities of a data center. Where the device power usage isn’t the only power consumption of the location, cooling must be taken into consideration as well. Redundant power switching also comes with a percentage loss in transmission efficiency which adds to power consumption and heat dispersion requirements.
It’s true, I don’t know how large the models are that are being accessed in data centers. Although if the article’s estimate is correct, it’s sad that such excessively-demanding models are always being used for use-cases that could often be handled with much lower power usage.
This seems a big waste of energy if that 140Wh (504,000 joules) number is correct. That amount of energy is about 2,000 times what it would take to do a very similar thing on a home PC.
Writing a 100 word email with a 7B model would take my PC about 5 seconds, times an increased power use of 50 watts, so 250 joules.
I get that they might be using a much larger model, but the e-mail is not going to be 2,000 times better.
That’s what I always thought when reading this and other articles about the estimated power consumption of GPT-4. Run a decent 7B LLM on consumer hardware like the steam deck and you got your e-mail in a minute with the fans barely spinning up.
Then I read that GPT-4 is supposedly a 1760B model. (https://en.m.wikipedia.org/wiki/GPT-4#Background) I don’t know how energy usage would scale with model size exactly, but I’d consider it plausible that we are talking orders of magnitude above the typical local LLM.
considering that the email by the local LLM will be good enough 99% of the time, GPT may just be horribly inefficient, in order to score higher in some synthetic benchmarks?
Computational demands scale aggressively with model size.
And if you want a response back in a reasonable amount of time you’re burning a ton of power to do so. These models are not fast at all.
Thanks for confirming my suspicion.
So, the whole debate about “environmental impact of AI” is not about generative AI as such at all. Really comes down to people using disproportionally large models for simple tasks that could be done just as well by smaller ones, run locally. Or worse yet, asking a behemoth model like GPT-4 about something that could and should have been a simple search engine query, which I (subjectively) feel has become a trend in everyday tech usage…
It’s about generative AI as it is currently used.
But yeah, the complaints everyone has about Gen AI are mostly driven by speculative venture capital. The only advantage Google and openai can maintain over open source models is a willingness to spend more per token than a hobbyist. So they’re pumping cash in to subsidize their LLMs and it carries with it a stupidly high environmental cost.
There’s no possible end game here. Unlike the normal tech monopolies, you can’t put hobbiest models out of business, by subsidizing your own products. But the market is irrational and expects a general AI, and is encouraging this behavior.
An article that thinks cooling is “consuming” should probably be questioned in all its claims.
I think there’s probably something wrong with the math around per-response water consumption, but it is true that evaporative cooling consumes potable water, in that the water cannot be reused until it cycles through the atmosphere and is recaptured from precipitation, same way you consume water by drinking and pissing it out, or agriculture consumes it for growing things. Fresh water usage is a major concern and bottleneck, especially with climate change. With the average data centre using 300k gallons of water per day, and Google’s entire portfolio using 5bn gallons per day, it’s not nothing.
The study that suggests 10-50 interactions with ChatGPT evaporates a whole bottle of water, doesn’t account for the fact that cooling systems are enclosed…
…and that “study” is based on a bunch of assumptions, which include evaporation from local power plants, as well as the entire buildings GPT’s servers are located in. It does this as if one user is served at a time, and the organizations involved (such as microsoft) do nothing BUT serve one use at a time. So the “study” (which isn’t peer reviewed and never got published) pretends those buildings don’t also serve bing, or windows, or all the other functions microsoft is involved with. It instead assumes whole buildings at microsoft are dedicated to serving just one user of ChatGPT at a time.
It also includes the manufacture of all the serve and graphics cards equipment, even though the former was used before ChatGPT, and will be used for other things as well… and the latter is only used in training.
You can check the study out yourself here:
http://arxiv.org/pdf/2304.03271
It’s completely junk. Worthless. Even uses a click bait title, and keeps talking about “the secret water foot print” as if it’s uncovering some conspiracy. It’s bunk science.
P.S It also doesn’t seem to understand that the bulk of GPT’s training was a one time cost, paid in 2021, with one smaller update in 2023.
I would say a model like ChatGPT could use a bit more energy than 7B llama
Datacenter LLM tranches are 7-8 H100s per user at full load which is around 4 kW.
Multiply that by generation time and you get your energy used. Say it takes 62 seconds to write an essay (a highly conservative figure).
That’s 68.8 Wh, so you’re right.
Source: I’m an AI enthusiast
Well that’s of the same order of magnitude as the quoted figure. I was suggesting that it sounded vastly larger than it should be.
They’re probably factoring in cooling costs and a bunch of other overhead, I dunno
Does that account for cooling? Storage? Networking? Non-H100 compute and memory?
Nope. Just GPU board power draw. 60 seconds is also pretty long with how fast these enterprise cards are but I’m assuming they’re using a giant 450B or 1270B model.
kW is a unit of instantaneous power; kW/s makes no sense. Note how multiplying that by seconds would cancel time out and return you power again instead of energy. You got there in the end, though.
Woop, noted, thanks
I like that the 140Wh is the part you decided to question, not the “consumes 1 x 500ml bottle of water”
That was covered pretty well already!
Or maybe it’s using Fluidic logic.
I’m sure I’m missing out, but i have no interest in using chatbots and other LLMs etc. It floors me to see how much attention they get though, how much resources are being dumped into their development and use. Nuclear plants being reopened for the sake of AI?!!
I also assume there’s a lot of things they’re capable of that could be huge for science, and there’s likely lots of big things happening behind closed doors that we’re yet to see in the coming years. I know it’s not all just chatbots.
The way this article strikes me though, is that it’s pretty much just wasting resources for parlor-game level output. I don’t know if i like the idea of people giving up their ability to write a basic letter or essay, not that my opinion on the matter is gonna change anything obviously 😅
Think of it like this: rich people accumulate more wealth by paying fewer people to accomplish more work faster, so it’s worth burning through the worlds resources at breakneck speed to help the richies out, right?
Nuclear plants being reopened for the sake of AI?!!
Do you have ANY evidence this is happening?
Oh, come on… it’s everywhere in the news for several months now, because all of the Big Techs suddenly (!) want to do that.
We live in a techno feudalism, it’s not a democracy, it’s no longer Capitalism. It’s Capitalism almost completely mediated through elite technocrats. The only freedom we have, the only potential for regaining control, is boycotting via massive, well organized, well understood, well conducted unsubscription programs, and campaigns. Lightening fast protest statements where we control who we fund, which services we use, and which we deny and defund by taking our usage data away (and their services away from us which takes discipline). Denying them data to sell, and profits/subscriptions.
This is the future politics we have to start discussing with others, the idea we have to spread, as it’s currently the only way to have a say within this current state of techno feudalism.
If we don’t we will end up in subscription model nations, living as slaves, with end user agreements, rather than human rights.
Mark my words: generative “AI” is the tech bubble of all tech bubbles.
It’s an infinite supply of “content” in a world of finite demand. While fast, it is incredibly inefficient at creating anything, often including things with dubious quality at best. And finally, there seems to be very little consumer interest in paid-for, commercial generative AI services. A niche group of people are happy to use generative AI while it’s available for free, but once companies start charging for access to services and datasets, the number of people who are interested in paying for it will obviously be significantly smaller.
Last I checked there was more than a TRILLION dollars of investment into generative AI across the US economy, with practically zero evidence of genuinely profitable business models that could ever lead to any return on investment. The entire thing is a giant money pit, and I don’t see any way in which someone doesn’t get left holding the $1,000,000,000,000 generative AI bag.
Don’t worry, we’ll bail them out once the bubble bursts.
I have read the comments here and all I understand from my small brain is that, because we are using bigger models which are online, for simple tasks, this huge unnecessary power consumption is happening.
So, can the on-device NPUs we are getting on flagship mobile phones solve these problems, as we can do most of those simple tasks offline on-device?
I’ve run an LLM on my desktop GPU and gotten decent results, albeit not nearly as good as what ChatGPT will get you.
Probably used less than 0.1Wh per response.
Is this for inferencing only? Do you include training?
Training is a one time thing. Tge more it get use, the less energy per query it will take
Good point. But considering the frequent retraining, the environmental impacts can only be spread on a finite number of queries.
They have already reached diminishing returns on training. It will become much less frequent soon. Retraining on the same data if there isn’t a better method is useless. I think the ressources consumed per query should only include those actually used for inference. The rest can be dismissed as bad faith argumentation.
Inference only. I’m looking into doing some fine tuning. Training from scratch is another story.
Yes, kind of… when those businesses making money out of the subscriptions are willing to ship with the OS for free which something only Apple has the luxury to do instead of OpenAI who doesn’t ship hardware or software (like Windows) beyond an app that’s less than 100MB. Servers would still be needed but not for general cases like help me solve this math or translation. Stable Diffusion or Flux is one example where you only need the connection to internet when downloading a certain model like you wouldn’t necessarily want to download every kind of game in the world when the intention is to play games arises.
🥵🥵🥵🔥🔥🔥💦💦💦
Why does the article make it sound like cooling a data center results in constant water loss? Is this not a closed loop system?
I’m imagining a giant reservoir heat sink that runs throughout a complex to pull heat out of the surrounding environment where some liquid evaporates and needs to be replenished. But first of all we have more efficient liquid coolants, and second that would be a very lazy solution.
I wonder if they’ve considered geothermal for new data centers. You can run a geothermal loop in reverse and use the earth as a giant heat sink. It’s not water in the loop, it’s refrigerant, and it only needs to be replaced when you find the efficiency dropping, which can take decades.
Yes, the vast majority are closed loop systems and the water isn’t really used up, like a lot of these headlines imply.
That’s not to say the energy being used can’t be put to better uses, though.
Not used up per se but sequestered. It’s water that nobody will ever get to drink or use for crops, etc.
The math on this doesn’t really check out. The USA uses 322 billion gallons of fresh water per day. A hyperscale datacenter uses only 5 million gallons per day.
There are about 1,000 hyperscale datacenters in the USA, so that comes out to 5 billion gallons of water every day.
That’s 1.5% of our annual freshwater usage, half of which is in closed loop systems and not going anywhere, and the other half being returned to the atmosphere where it will rain back down as fresh water again.
And of course, the water cycle doesn’t really care about national borders or annual evaporation rates so much, and there is about 1 quintillion gallons of liquid fresh water available worldwide, so its not like sequestering 5 million gallons really offsets the available freshwater needed for hydration and agriculture.
It could be used for other things like district heating at least.
Datacenters are usually not located where this would be useful. They’re placed where space and energy are cheap, because everything they do only needs Internet access. At most they’d heat the rest of the building for whatever office space there is.
It is a closed loop, but the paper treats it as if it’s an open loop, and counts all water use for the building, as well as all the water that went into creating any equipment used… and the water that escapes power plants in powering the buildings… it also includes any other buildings that might house related services. Here is the original “study” which is about what maths could be done given the above assumptions:
http://arxiv.org/pdf/2304.03271
In short, it has nothing to do with reality, and is more just an attempt at the authors to get their names out there (on bad science that the media is interested in publicizing for click bait reasons).
Evaporative coolers save a ton of energy compared to refrigerator cycle closed loop systems. Like a swamp cooler, the hot liquid that comes from cooling the server is exposed to the atmosphere and enough evaporates off to cool the liquid by a decent percentage, then it’s refrigerated before going back into the servers.
Data centre near me is using it and the fire service is used to be being called by people concerned the huge clouds of water vapor are smoke
It highly depends on every data center, but it is very likely that they do use municipal water for cooling. Mainting a Reservoir is extremely expensive for the amount of thermal mass it requires, these things kick off HEAT.
I don’t know why they aren’t using reclaimed water from treatment plants. I don’t see why potable water is necessary as long as the substitute isn’t corrosive, but I might be missing something here.
You’d have to get the gray water in, and it’s more efficient to just continue treating it and using the municipal water system.
You can run a geothermal loop in reverse and use the earth as a giant heat sink.
You need something to move the heat away, like water or air. Having something solid that just absorbs will reach its heat capacity pretty quick.
deleted by creator
Deep Geothermal goes deeeeeepppp to where there is a heat source that is replenished.
Shallow geothermal pulls heat where there is no replenishment, and you have to run it in reverse (use it as an AC in the summer) to swap out the heat. You can’t only pull heat out for shallow geothermal. You may be able to for a time, but also remember that heating for a house is pretty small overall.
It’s not the entire earth that is the heat sink, it’s a relatively short distance from the pipe. We don’t get the massive heat from the molten core at the surface.
Can we PLEASE shut that shit down? We were doing just fine without it.
Billions of people have been using Google for years, and Google has been using artificial intelligence for years. It’s nothing new.
How do you type “it’s nothing new” about a burgeoning new industry and take yourself seriously
I don’t understand the hate for AI. It’s a new technology that has some teething issues, but it’s only going to get better and more efficient.
it won’t if you don’t force it to. that’s like saying companies will pollute less if you give them enough time. no, you have to grab their balls and force them to do it.
I think it’s fair to say that pretty much every industry is more efficient and cleaner than it used to be and I don’t see why AI would be an exception to that.
And why do you think those improvements happen?
Is it (a) unchecked capitalism or (b) regulations?
Mainly because energy and data centers are both expensive and companies want to use as little as possible of both - especially on the energy side. OpenAI isn’t exactly profitable. There is a reason companies like Microsoft release smaller models like Phi-2 that can be run on individual devices rather than data centers.
I didn’t realize coal plants were concerned about data centers or AI. TIL.
But in the interest of being slightly less of a dick and responding to what you said even though it’s kinda a non sequitur, companies are only vaguely interested in efficiency. I think it’s more accurate to say that AI is hot for everyone right now so there’s more eyes on it which makes the concept you laid out valid. Where it’s invalid in my experience is that efficiency is just based on “where x executive is paying attention” not an honest attempt to look at return on investment in a rigorous way across the enterprise.
I didn’t realize coal plants were concerned about data centers or AI. TIL.
What? How does that relate to anything I just said?
But in the interest of being slightly less of a dick and responding to what you said even though it’s kinda a non sequitur, companies are only vaguely interested in efficiency.
How is it a non sequitur? If anything the thing you just said makes no sense. Energy is probably the biggest cost these companies have. This I believe is true even for regular data centers and cloud services which is why they always try to use the latest most energy efficient hardware. It’s still not as bad as most anti-AI people seem to believe, mainly because the most energy intensive part happens only once per model (training).
I think it’s more accurate to say that AI is hot for everyone right now so there’s more eyes on it which makes the concept you laid out valid. Where it’s invalid in my experience is that efficiency is just based on “where x executive is paying attention” not an honest attempt to look at return on investment in a rigorous way across the enterprise.
Human labour is expensive. So trying to replace it with AI, even if AI is also expensive, is typically still worth it.
You talk about experience, but I honestly don’t think you have any. Do you actually work in tech? What are your qualifications? Most of the people coming here to complain about this stuff don’t actually have a functional understanding of the thing they are complaining about.
Is the insinuation here that the AI industry is unregulated? Because I’m not against regulations that would drive these improvements.
No? What are you talking about?
Well your comment sounded like you were implying that regulations are needed that currently aren’t there.
i think you’re not thinking about what efficiency means for corporations.
I think it’s exactly what I’m thinking about, unless I’m missing something specific that you’d like to put forward?
If I own a bottled drinks company and the energy cost is 10p a bottle but a new, more efficient process is invented that would lower my energy cost to 5p a bottle, that’s going to be looking like a wise investment to make. A few pence over several thousand products adds up pretty quickly.
I could either pocket the difference as extra profit, lower my unit price to the consumer to make my product more competitive in the market, or a bit of both.
It seems basic logic like this doesn’t actually work on these people. Most really can’t get their heads around the fact that energy costs money and companies want to use less of it wherever possible and practical to do so.
The mob is fickle, brother.
The hive mind has concluded AI=bad and any comment that doesn’t go along with the consensus is going to get downvotes.
It’s really not that different from the beginning of the industrial revolution, when cotton mills first started to implement the spinning jenny, leaving many workers out of a job who’d go in to the factories at night to smash the machines up.
No one wants to go back to spinning cotton all day now though and it will be the same with jobs taken by AI.
Until it does, we shouldn’t exacerbate the climate and resource issues we already have by blindly buying into the hype and building more and larger corporate-scale power gluttons to produce even more heat than we’re already dealing with.
“AI” has potential, ideas like machine assistance with writing letters and improving security by augmenting human alertness are all nice. Unfortunately, it also has destructive potential for things like surveillance, even deadlier weapons or accelerating the wealth extraction of those with the capital to invest in building aforementioned power gluttons.
Additionally, it risks misuse and overreliance, which is particularly dangerous in the current stage where it can’t entirely replace humans (yet), the issues of which may not immediately become apparent until they do damage.
If and until the abilities of AI reach the point where they can compensate tech illiteracy and we no longer need to worry about the exorbitant heat production, it shouldn’t be deployed at scale at all, and even then its use needs to be scrutinised, regulated and that regulation is appropriately enforced (which basically requires significant social and political change, so good luck).
If and until the abilities of AI reach the point where they can compensate tech illiteracy and we no longer need to worry about the exorbitant heat production, it shouldn’t be deployed at scale at all, and even then its use needs to be scrutinised, regulated and that regulation is appropriately enforced (which basically requires significant social and political change, so good luck).
Why wouldn’t you deploy that kind of AI at scale?
To be honest I think people keep forgetting that AI strong enough would be smarter than a human, and would probably end up deploying us at scale rather than the other way around. Terminator could one day actually happen. I am not even sure that would be a bad thing given how flawed humans are.
AI strong enough would be smarter than a human
General AI might be, but the type of “AI” we have right now isn’t general, isn’t smarter, it’s just a really expensive imitation engine that people keep mistaking for actual intelligence.
And the energy consumption and heat production are really not what our global situation needs right now.
deleted by creator
This is the most pedantic reply so far.
yeah, but it can do really cool things like “suggest a name for my project that does X”.
surely that game’s worth the candle, yes?
Hah! Haha! Hahahaahah! Ties well with this one news article that I glimpsed that claims that by 2030 the need for fresh water will be 140% of the world’s freshwater reserves. Infinite growth forever!
Time to buy stock in water lol
So, Nestlé stocks?
New admin will do its part by discouraging pregnancy and encouraging people to die sooner.
Absolutely my favorite game ever
Blowing stuff up with Arty is never not funny for me
The real surprise for me is how little the battery of my iphone holds. Especially compared to my ev6 or what my heat pump guzzles daily. Crazy.
