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
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!
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 😅
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.
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…
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.
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.
Does that account for cooling? Storage? Networking? Non-H100 compute and memory?
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.
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.
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.
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.
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.
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 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).
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 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.
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.
