Despite its name, the infrastructure used by the “cloud” accounts for more global greenhouse emissions than commercial flights. In 2018, for instance, the 5bn YouTube hits for the viral song Despacito used the same amount of energy it would take to heat 40,000 US homes annually.
Large language models such as ChatGPT are some of the most energy-guzzling technologies of all. Research suggests, for instance, that about 700,000 litres of water could have been used to cool the machines that trained ChatGPT-3 at Microsoft’s data facilities.
Additionally, as these companies aim to reduce their reliance on fossil fuels, they may opt to base their datacentres in regions with cheaper electricity, such as the southern US, potentially exacerbating water consumption issues in drier parts of the world.
Furthermore, while minerals such as lithium and cobalt are most commonly associated with batteries in the motor sector, they are also crucial for the batteries used in datacentres. The extraction process often involves significant water usage and can lead to pollution, undermining water security. The extraction of these minerals are also often linked to human rights violations and poor labour standards. Trying to achieve one climate goal of limiting our dependence on fossil fuels can compromise another goal, of ensuring everyone has a safe and accessible water supply.
Moreover, when significant energy resources are allocated to tech-related endeavours, it can lead to energy shortages for essential needs such as residential power supply. Recent data from the UK shows that the country’s outdated electricity network is holding back affordable housing projects.
In other words, policy needs to be designed not to pick sectors or technologies as “winners”, but to pick the willing by providing support that is conditional on companies moving in the right direction. Making disclosure of environmental practices and impacts a condition for government support could ensure greater transparency and accountability.
This is the best summary I could come up with:
It is hardly news that the tech bubble’s self-glorification has obscured the uglier sides of this industry, from its proclivity for tax avoidance to its invasion of privacy and exploitation of our attention span.
The industry’s environmental impact is a key issue, yet the companies that produce such models have stayed remarkably quiet about the amount of energy they consume – probably because they don’t want to spark our concern.
Google’s global datacentre and Meta’s ambitious plans for a new AI Research SuperCluster (RSC) further underscore the industry’s energy-intensive nature, raising concerns that these facilities could significantly increase energy consumption.
Additionally, as these companies aim to reduce their reliance on fossil fuels, they may opt to base their datacentres in regions with cheaper electricity, such as the southern US, potentially exacerbating water consumption issues in drier parts of the world.
In an era where we expect businesses to do more than just make profits for their shareholders, governments need to evaluate the organisations they fund and partner with, based on whether their actions will result in concrete successes for people and the planet.
As climate scientists anticipate that global heating will exceed the 1.5C target, it’s time we approach today’s grand challenges systemically, so that the solution to one problem does not exacerbate another.
The original article contains 766 words, the summary contains 214 words. Saved 72%. I’m a bot and I’m open source!
Yes it does, and wait until you hear about literally every other industry.
“The world is complicated and scary! I don’t understand it so it must be bad! M-muh planet farting cows evil industry fuck the disabled/sick/queer!” - What luddites actually believe.
Anprims/eco-fashes begone. If the planet was destroyed for the betterment of conditions for the proletariat today and future alike there’d be literally no issue, it’s just some rock lol, AI is far more important. Also brutalism and soviet blocs are the best architectural styles, everything else is bourgeois cringe.
That’s wrong, I buy drugs online with cryptocurrencies all the time to this day and have done it long before the normies showed up and turned it into a mostly financial scam.
Evading the man and LEOs when the law ain’t right is my god-given right and I’m thankful to be born in the age of onions and crypto.
“aI AnD cRyPtO aRe ThE sAmE bRo”
You know that your take that they both must suck in the exact same ways just because tech bros get hyped about them, is literally just as shallow, surface level, and uninformed as most tech bros?
Like yeah man, tech hype cycles suck. But you know what else was once a tech hype cycle? Computers, the internet, smartphones. Sometimes they are legitimate, sometimes not.
AI is solving an entirely new class of problem that computers have been literally unable to solve for their entire existence. Crypto was solving the problem of making a database without a single admin. One of those is a lot more important and foundational than the other.
On top of that, crypto algorithms are fundamentally based on “proof of work”, i.e. literally wasting more energy than other miners in the network is a fundamental part of how their algorithm functions. Meaning that with crypto there is basically no value prop to society and it inherently tries to waste energy, neither is the case for AI.
Plus guess how much energy everyone streaming 4K video would take if we were all doing it on CPUs and unoptimized GPUs?
Orders of magnitude more power than every AI model put together.
But guess what? Instead we invented 4k decoding chips that are optimized to redner 4k signals at the hardware level so that they don’t use much power, and now every $30 fire stick can decode a 4k signal on a 5V usb power supply.
That’s also where we’re at with the first Neural Processing Units only just hitting the market now.
Cryptos have drastically reduced their energy consumption through technological improvements.
That’s why nobody complains about crypto energy consumption anymore. It’s just bitcoin.
But these LLMs just need more and more with no end in sight.
Funny how 99.99% of cryptos shrivel up and die while bitcoin continues to serve people all over the world and is constantly becoming more and more popular. Maybe if you lived with, or even gave a shit about, people in below average wealth countries you would understand why Bitcoin is so useful to them.
Go on benefiting from the people who actually do stuff while simultaneously whining about it. You’ve been using AI for 20 years, you’re just too thick to know about it. There are millions of people in 2nd and 3rd world countries who have had their lives massively improved thanks to bitcoin, you’re just too spoiled and naive and to give a shit about them. Climb down off your soap box and go read something beyond the headline.
That other poster is using a disingenuous debate tactic called “whataboutism”. Basically shifting the focus from what’s being criticised (AI resource consumption) to something else (other industries).
Your comparison with evangelists is spot on. In my teen years I used to debate with creationists quite a bit; they were always
- oversimplifying complex matters
- showing blatant lack of reading comprehension, and distorting/lying what others say
- vomiting certainty on things that they assumed, and re-eating their own vomit
- showing complete inability to take context into account when interpreting what others say
- chain-gunning fallacies
- “I’m not religious, but…”
always to back up something as idiotic as “the world is 6kyo! Evolution is a lie!”.
Does it ring any bell for people who discuss with AI evangelists? For me, all of them.
(Sorry bolexforsoup for the tone - it is not geared towards you.)
So… Absolutely need to be aware of the impact of what we do in the tech sphere, but there’s a few things in the article that give me pause:
Research suggests, for instance, that about 700,000 litres of water could have been used to cool the machines that trained ChatGPT-3 at Microsoft’s data facilities.
- “Could”. More likely it was closed loop.
- Water isn’t single use, so even if true how does this big number matter.
What matter is the electrical energy converted to heat. How much was it and where did that heat go?
Moreover, when significant energy resources are allocated to tech-related endeavours, it can lead to energy shortages for essential needs such as residential power supply. Recent data from the UK shows that the country’s outdated electricity network is holding back affordable housing projects.
Can you say non sequitur ?
The outdated network holding back housing is that it doesn’t go to the right places with the capacity needed for the houses. Not that OpenAIUK is consuming so much that there’s no power left. To use a simily, there’s plenty of water but the pipes aren’t in place.
This article is well intentioned FUD, but FUD none the less.
700.000 litres also sounds like much more than 700 m³. The average German citizen consumed 129 litres per day or roughly 47 m³ annually. The water consumption of 15 people is less than most blocks.
Energy consumption might be a real problem, but I don’t see how water consumption is that big of a problem or priority here.
The average German citizen consumed 129 litres per day
That seems like a lot. Where are you getting that number?
Edit: consumes = uses not drinks
A quick search says 3.7L is the recommended intake for men, and 2.7L for women. Forget AI, Germans appear to be the real resource guzzlers!
The EPA states that each American uses an average of 82 gallons or 310.4 litres a day (study from 2015). Source: https://www.epa.gov/watersense/statistics-and-facts
Liters are a great unit for making small things seem large. I’ve seen articles breathlessly talking about how “almost 2000 liters of oil was spilled!” When 2000 liters could fit in the back of a pickup truck.
Water “consumption” is also a pretty easy to abuse term since water isn’t really consumed, it can be recycled endlessly. Whether some particular water use is problematic depends very much on the local demands on the water system, and that can be accounted for quite simply by market means - charge data centers money for their water usage and they’ll naturally move to where there’s plenty of cheap water.
Oil is different because 1 ppm can ruin a whole litre or something in that direction.
Liters are a great unit for making small things seem large. I’ve seen articles breathlessly talking about how “almost 2000 liters of oil was spilled!” When 2000 liters could fit in the back of a pickup truck.
That just means you have no intuitive sense of how large a litre is. If they’d written it as “2000 quarts” (which is close enough to being the same volume at that level of rounding) would it have painted a clearer picture in your head?
“Could”. More likely it was closed loop. As I understand it this is an estimate, thus the word “could”. This has nothing to do with using closed or open look water cooling. Water isn’t single use, so even if true how does this big number matter.
The point they are trying to make is that fresh water is not a limitless resource and increasing usage has various impacts, for example on market prices.
The outdated network holding back housing is that it doesn’t go to the right places with the capacity needed for the houses. Not that OpenAIUK is consuming so much that there’s no power left. To use a simily, there’s plenty of water but the pipes aren’t in place.
The point being made is that resources are allocated to increase network capacity for hyped tech and not for current, more pressing needs.
Is there a reason it needs to be fresh water? Is sea water less effective?
A lot of industry does use grey water or untreated water for cooling as it’s substantially cheaper to filter it and add chemicals to it yourself. What’s even cheaper is to have a cooling tower and reuse your water, in the volumes it’s used at industrial scales it’s really expensive to just dump down the drain (which you also get charged for), when I worked as a maintenance engineer I recall saving something like 1m cad minimum a year by changing the fill level in our cooling tower as it would drop to a level where it’d trigger city water backups to top up the levels to avoid running dry, and that was a single processing line.
“Could”. More likely it was closed loop.
Nope. Here’s how data centres use water.
It boils down to two things - cooling and humidification. Humidification is clearly not a closed loop, so I’ll focus on the cooling:
- cold water runs through tubes, chilling the air inside the data centre
- the water is now hot
- hot water is exposed to outside air, some evaporates, the leftover is colder and reused.
Since some evaporates you’ll need to put more water into the system. And there’s an additional problem: salts don’t evaporate, they concentrate over time, precipitate, and clog your pipes. Since you don’t want this you’ll eventually need to flush it all out. And it also means that you can’t simply use seawater for that, it needs to be freshwater.
Water isn’t single use, so even if true how does this big number matter.
Freshwater renews at a limited rate.
What matter is the electrical energy converted to heat. How much was it and where did that heat go?
Mostly to the air, as promoting the evaporation of the water.
Can you say non sequitur ?
More like non sequere than non sequitur. Read the whole paragraph:
Moreover, when significant energy resources are allocated to tech-related endeavours, it can lead to energy shortages for essential needs such as residential power supply. Recent data from the UK shows that the country’s outdated electricity network is holding back affordable housing projects. This will only get worse as households move away from using fossil fuels and rely more on electricity, putting even more pressure on the National Grid. In Bicester, for instance, plans to build 7,000 new homes were paused because the electricity network didn’t have enough capacity.
The author is highlighting that electrical security is already bad for you Brits, for structural reasons; it’ll probably get worse due to increased household consumption; and with big tech consuming it, it’ll get even worse.
Data center cooling towers can be closed- or open-loop, and even operate in a hybrid mode depending on demand and air temps/humidity. Problem is, the places where open-loop evaporative cooling works best are arid, low-humidity regions where water is a scarce resource to start.
On the other hand, several of the FAANGS are building datacenters right now in my area, where we’re in the watershed of the largest river in the country, it’s regularly humid and rainy, any water used in a given process is either treated and released back into the river, or fairly quickly condenses back out of the atmosphere in the form of rain somewhere a few hundred miles further east (where it will eventually collect back into the same river). The only way that water is “wasted” in this environment has to do with the resources used to treat and distribute it. However, because it’s often hot and humid around here, open loop cooling isn’t as effective, and it’s more common to see closed-loop systems.
Bottom line, though, I think the siting of water-intensive industries in water-poor parts of the country is a governmental failure, first and foremost. States like Arizona in particular have a long history of planning as though they aren’t in a dry desert that has to share its only renewable water resource with two other states, and offering utility incentives to potential employers that treat that resource as if it’s infinite. A government that was focused on the long-term viability of the state as a place to live rather than on short-term wins that politicians can campaign on wouldn’t be making those concessions.
They can be closed-loop as in your region but they usually aren’t - besides the problem that you mentioned, a closed loop increases electricity consumption (as you’ll need a heat pump instead), and electricity consumption is also a concern. Not for the environmental impact (corporations DGAF), but price.
Despite its name, the infrastructure used by the “cloud” accounts for more global greenhouse emissions than commercial flights. In 2018, for instance, the 5bn YouTube hits for the viral song Despacito used the same amount of energy it would take to heat 40,000 US homes annually.
Mixing and matching abstract measurements doesn’t work when comparing two things.
it actually is an enlightening comparison when you dig into it. It’s saying that the energy required to power one play of a song is 4e4*365/5e9 of the energy to heat a home for one day. That comes out to about 0.3%, i.e. if you watch a three minute youtube video three times and do absolutely nothing else that day but heat your house (dont use any other electricity, dont eat anything, dont travel anywhere) you increase your energy usage by a total of 1%
I can’t tell if this is serious since most homes don’t need heated every day…
Yeah thats bullshit. Unless you have a hyper efficient heating system and power your internet with a badly tuned 1950s generator, theres no way youre getting 1%.
It does not work like that.
The problem with such statements is the energy costs are nowhere near fixed. The amount of energy needed to play a song on my iPod shuffle through a wired headset is wildly different from the power needed to play that same song on my TV through my home theater equipment.
The same is true on the backend. The amount of power Google spends serving up a wildly popular band is way less than what they burn serving up an unknown Indy band’s video. That’s because the popular band’s music will have been pre-optimized by Google to save on bandwidth and computing resources. When something is popular, it’s in their best interests to reduce the computational costs (ie power consumption) associated with serving that content.
It’s a new blockchain. It’ll fizzle out but we’ll come up with a new buzzword by then.
It won’t fizzle out; it already has legitimate business use cases. (A lot fewer than the marketing bros want you to believe, but real use cases nonetheless.) Blockchain and Augmented Reality never reached this point, so they fizzled. We’ll see a huge AI winter soon just like we did in the dot com bust in 2000.
Arguably that may be related - cryptocurrency people needed a new thing to prop up their Nvidia shares, and “AI” fills that niche.
It will not be economically viable once AI companies have to pay for their training data. So far they made some deals with press/media but multimedia is a can of worms that’s waiting to explode in our faces. They’re getting away with this because doing things and then asking for permission / forgiveness is a very Sillicon Valley thing to do, for now.
Technology itself seems to be in a plateau. The whole AI computer thing is just moving computation offline because amounts of energy needed are unsustainable and have to be dumped on consumers. We haven’t seen that much progress since ChatGPT took the world by storm.
I’m not saying AI is a fad. It’s revolutionizing medical research for example, and those industries actually own the data they’re training AI on. EU sees this and is currently working on streamlining exchanging this data across member states too.
thing is that few if any use cases for blockchain were found and any actual useful things would not require much energy. The high energy crypto itself does nothing useful over more efficient alternatives and I don’t know what you mean by fizzle out but it still uses massive amounts of energy. the language models unfortunately do things that are useful and is much more likely to keep drawing power.
And the really perverted incentive of crypto is that due to the way difficulty is done, in particular with PoW systems, the more adoption there is the more energy intensive it becomes. Scaling actually leads to more inefficiency by design. I mean it’s totally asinine.
oh yeah. in the end you have a system that creates artificial value by requiring the sacrifice of real value. heres one credit for burning a barrel of oil. oh now you have to burn 2 to get a credit, now its 4, now its 8.
