I'm glad somebody out there is brave enough to push back against the "personal ChatGPT usage is terrible for the environment" message https://andymasley.substack.com/p/a-cheat-sheet-for-conversations-about
"If you want to prompt ChatGPT 40 times, you can just stop your shower 1 second early."
"If I choose not to take a flight to Europe, I save 3,500,000 ChatGPT searches. this is like stopping more than 7 people from searching ChatGPT for their entire lives."
This is a recasting of the "it's ridiculous to ban private jets - they are only a small percentage of CO2 emissions" argument.
We consume prodigious amounts of the Earth's resources. We need to stop encouraging even more consumption.
If someone needs to loose weight to get healthy, do we offer a dessert mint after the burger they already eat?
Be ashamed of bad habits; don't use them to justify new ones.
That's why I got rid of my car, stopped flying, use green energy, etc.
Just a small note, though: using weight or eating habits as a comparison can unintentionally come across as fat-shaming. We can discuss better choices for the planet without making people feel bad about their bodies. (I'm going to assume you meant well here.)
Outside of the drive-by reply, I didn't realize how much online streaming's environmental footprint was.
The flip side is when people who need to lose weight for health reasons get shamed as fat-phobic.
(Eg, https://medium.com/in-fitness-and-in-health/shamed-as-fatphobic-for-pursuing-better-health-d530ad30b1d behind a Medium wall but available elsewhere.)
On the other hand, I want to make flyskam - "flight shame" - real and widespread.
Discussing things nicely hasn't worked during the last 35 years.
Re: online streaming, I wonder how much of that is from the media transfer, and how much for the ad system (user tracking, customized bids, etc.)
In my journey to improve my health and wellness, I’ve gotten lost within the dark side of the fat acceptance, body positivity, and Health At Every Size (HAES) communities. Hi, I’m Holly, and I’m a…
My point is to look at one's entire CO2/pollution budget, not make relative comparisons.
Dickens: "Annual income £20, annual expenditure £19/6, result happiness. Annual income £20 pounds, annual expenditure £20/0/6, result misery."
Collectively exceeding our budget will bring misery.
Two flights per year might be within the CO2 budget, in which case - happiness!
OTOH, two flights + ChatGPT, might bust the budget - misery.
@simon It's almost hard to believe "green" distractions like this aren't intended to prevent more effective actions (e.g. spending less money).
Also worth noting that Google Gemini is probably 80% more efficient: https://venturebeat.com/ai/the-new-ai-calculus-googles-80-cost-edge-vs-openais-ecosystem/
@passwordsarehard4 @Ulli you are making a slightly different argument there
The piece argues that it's OK spending minimal energy on things that are useless (which the author and myself both believe or to be the case for LLMs)
It looks to me like you are arguing against spending energy on things that are actively harmful
@passwordsarehard4 @Ulli and yes, if there were no way to use LLMs that did not actively harm the user I would support discouraging their use of even outright banning them, independently of their energy usage
I do not believe it is the case that all uses of LLMs actively harm their users - I think they require thoughtful application and we need to work hard to help people understand their many limitations and flaws
@Ulli one of the most important skills for making effective use of LLMs for coding assurance is being *really good* at code review
Engineers with great code review skills (and who don't try to avoid reading other people's code because they prefer to write their own) can get a whole lot more value out of LLMs
@simon Using it promotes AI as a viable business model as a whole as well as (in the case of non-local models) provides usage data and training content (and money if you pay) for the companies to do further bad AI business with.
So of course, while the direct impact of one query on the environment doesn't really matter, the economical impact of you using it does indeed have an influence on the environment!
The goal has to be to make AI and its connected surveillance capitalism a non-viable business model, and refraining to use it and shunning everyone who does is a way to do that!
@simon
Give me an approach which:
and maybe then we can talk about actually deploying this in the real world. Until then they should stop their venture-capital induced hype bubble fueled by ex-crypto-bros und get back the the drawing board. I'll take human work over an AI every day until that happens...
@the_moep we have some of those today:
- Does not waste massive amounts of energy (and water) for training: DeepSeek v3, still one of the strongest models - trained for just $6m, way less than the biggest USA models
- Does not require expensive rare materials to run: can't help with that if it rules out laptops
- Doesn't spit out wrong crap: yeah that's still unsolved! Newer models are "better", but honestly an AI tool that never makes a mistake feels like it will always be science fiction me
- Can actually run locally in a private environment: yes! We have that now. The models I can run on my laptop got really good starting from about six months ago
- Is free (as in libre) to be used by everyone: we are there too. The Qwen models are under an Apache 2 open source license. Plenty of other good models are "open weights" which is almost good enough to allow "free to be used by everyone"
@simon
Pretty much all of that does not hold true in my opinion:
DeepSeek v3, still one of the strongest models - trained for just $6m, way less than the biggest USA models
That's money, not power or resources. Any monetary cost claimed by a Chinese company can't be compared to actual free countries as a way larger part of that cost is offset by reduced environmental and worker protection (and partial slavery), so totalitarian in general. Also they allegedly based their work on OpenAI so you might have to add their costs on top too...
Does not require expensive rare materials to run: can't help with that if it rules out laptops
It's about their requirement of specialized hardware to train (while the models might run on "normal" CPU nowadays, they cannot be trained on a cheap phone or laptop. A normal program can be created there no problem.
Can actually run locally in a private environment: yes! We have that now. The models I can run on my laptop got really good starting from about six months ago
They really aren't. They are slow and can't handle actual reasoning or even remembering things like the "big" models can. (And of course they are anything but intelligent.)
But it's all just a word prediction system anyways I guess it now just predicts more words at a basically cost linear to how much words you input and want to have predicted, so with the current approach a local machine will always be behind what one can do in a huge data-center hence why I want a different approach that aren't just llm and inference-based.
Is free (as in libre) to be used by everyone: we are there too. The Qwen models are under an Apache 2 open source license. Plenty of other good models are "open weights" which is almost good enough to allow "free to be used by everyone"
All they release is the finished model (and in the case of Qwen their weights) which of course is nice, but it does not allow reproducing or even forking their work. As long as they do not release the code which made the model and the training data under a free license it imo. cannot be considered free.
Them licensing it under any kind of free software license might actually not be valid as it's not based on work that was available under a free license. I would even go as far and say that most models are released illegally as they are derivatives of copyrighted works. Them sticking a free software license on it does not magically safe them from the copyright the material they used is under.
A good comparison in the classical software development work is the CraftBukkit project which used Mojang's Copyrighted Minecraft code and got taken down not by Mojang/Microsoft but by a contributor because their approach violated the GPLv3 license, most "open" models run into the same issue
@the_moep shunning people who use LLMs because they are being environmentally irresponsible feels dishonest to me
There are plenty of credible arguments against irresponsible usage of LLMs, I don't like seeing people waste their time on the ones that are least credible
@simon Well while I agree that one should use the correct arguments for criticizing something I also don't think one should completely discard the personal environmental-impact argument when it comes to usage of AI.
I see the issue with that argument more in that the real cost of AI-queries are hidden from the user (and independent researchers), it's not the energy spent on the single query but the amount of resources spent to create the model and to further the development of the current approach. (Which includes discarding existing hardware or potentially even models)
Of course that is way harder to calculate but I believe that you would get closer to other things that are now generally accepted as being bad for the environment like short-range flights.
@tasket You are of course right, sorry. It seems that I too fell victim to all the "AI" hype talk, but I feel the actual meaning of that term is now (unfortunately) generally accepted as meaning "generative AI, especially LLMs".
I do not have such a big issue with actual useful "AI" doing stuff like voice recognition or protein folding.
@simon actually it's not that this is good news. It's the aggregate value that is impressive. As it is 1 second of shower, when you multiply it by 7 billions. However i do recognize that there are considerations to be made. One of those is that roughly 1/7 of population today probably did not have even the chance for a shower. So, abruptly, a 1-second shower takes a whole new dimension, and 10 ChatGPT queries for each of the other 6 billions people mean a full 1 minute shower for the remaining 1. Probably this could prevent a few millions deaths by infection each year.
There are dozens of ways each one of us wastes resources that mean life or death to others, and I think no one with this awareness is happy when wasting. Simply we didn't need ChatGPT to be another one, as we didn't need any crypto or any metaverse.
Defending it and setting the stage for confrontation is, in the best frame I can think of, love for numbers and I can accept it. But however little, to me is however too much.
@Xzan the training cost estimates are covered here https://andymasley.substack.com/p/a-cheat-sheet-for-conversations-about#§training-an-ai-model-uses-too-much-energy (estimated 10% increase in cost-per prompt after amortization)
I've not seen anyone ever account for the cost of failed runs, which is frustrating
@trevorthetuba @simon I got tired of all the AI slop in a Github repo (the issues text, luckily others go through that particular code as it is JS/TS) and recommended this to my colleagues: https://gptzero.me/
It detects that the author is an LLM even if the LLM is told to deliberately include spelling and grammar mistakes. Must be based on the usage of big words (who says "paramount" IRL) and low information density.
I do not need that tool, I can detect AI slop just by how tiresome it is to read, I've only used it 10 or so times (for both LLM and human written texts) to confirm my hunches.
@simon I think a big problem with AI is that it's being forced upon us at a time when our polution should be decreasing, and should have been for many years.
And now, instead of it, we've figured out a new way to burn fossil fuels for something we never needed in the first place. (Or at best, marginally needed, I mean, we already had artists and notulists etc).
So I see this argument (of consumption by LLM) as a generic system critique. Just like we need to get rid of our plastic dependency, but if you measure it per person per packaged piece of candy: is super small.
@platlas the environmental impact of Bitcoin feels different to me because of the way that system is designed: as a competition
To mine new Bitcoin you don't just have to burn energy: you have to burn MORE energy than anyone else
It's impossible to make that system more efficient, because any efficiencies will be lost to the increased competition they cause between rival miners
Jonathan
Simon Willison
Jeff Triplett
Andrew Dalke
Sadie
Ulli
Djoerd Hiemstra 🍉
Scott J.
frew
Lars Marowsky-Brée 😷
Max Lee
"Wish the sun to stand still.." 🌞
TiTiNoNero 
Benjamin Monjoie
Mark Einon
Trevor Mullen
Jeffy 🏳️⚧️🏳️🌈🇺🇦 ❤️🇱🇺
Enola Knezevic
Solarpunk Davy
platlas