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."

Using ChatGPT is not bad for the environment - a cheat sheet

The numbers clearly show this is a pointless distraction for the climate movement

Andy Masley
Here's a follow-up by someone with a whole lot more credibility than myself or Andy: Hannah Ritchie is "Head of Research at Our World in Data, and a Senior Researcher at the University of Oxford" and runs a high quality sustainable energy blog https://www.sustainabilitybynumbers.com/p/carbon-footprint-chatgpt
What's the carbon footprint of using ChatGPT?

Very small compared to most of the other stuff you do.

Sustainability by numbers

@simon Thanks for sharing! That graph about half-way down is quite telling...

> …if we’re fretting over a few queries a day while having a beef burger for dinner, heating our homes with a gas boiler, and driving a petrol car, we will get nowhere.

@carlton @simon I see at least 6 from the "important" ones which can't be just delegated to the people choice (some are a "luxury" many people can't afford, some need government help etc...). And I'm not saying the chart is wrong. I'm saying that if true, we are getting nowhere, regardless of LLM 😕

@andreagrandi @simon AFAICT the consensus among activists who know about this sort of thing is that, you're right, we're not getting anywhere.

This was a good read:

https://www.lrb.co.uk/the-paper/v47/n02/brett-christophers/so-much-for-paris

As well as the pieces mentioned, the author there has a book "The Price Is Wrong" which outlines the dynamics, somewhat depressingly.

Brett Christophers · So much for Paris: Climate Overshoot

All fossil fuel projects are politically constituted. Either they are state projects, controlled and administered by...

London Review of Books
@carlton @simon ni tengo coche, ni como vacas y en casa todo es eléctrico así que voy a pasar por el momento de la brasa sobre el consumo de recursos de las IAs. Igual que pasé de los pajaritos muertos por las eólicas y de los cultivos arrasados por la fotovoltaica. Porque ya no sé si de verdad se lo creen o es todo propaganda nazi para despistar. Jamás hacen unos números, ni dan contexto, ni plantean alternativas viables, solo intentan asustar y culpabilizar y yo ya no les aguanto y además me parecen nocivos en su mayor parte. Tengo el TL un poco saturado, me vendrá bien dejar de seguirles y algún bloqueo caerá también.

@simon

> For the regular or even relatively high user of text-based LLMs: stop stressing about the energy and carbon footprint. It’s not a big deal, and restraining yourself from making 5 searches a day is not going to make a difference. In fact, it might have a net negative impact because you’re losing out on some of the benefits and efficiencies that come from these models.

👀 Noted.

@simon Counterpoint: the environmental impact of individuals' personal use of genAI ("here's my cute action figure!" etc) may be negligible, but in aggregate it arguably has a massive role to play in legitimising the current wave of data centre buildouts by hyperscalers, VCs etc. And for now [*] that's only possible through undermining or abandoning prior Net Zero commitments.

[*] Future visions of yeeting data centres into space notwithstanding: https://arstechnica.com/space/2025/05/eric-schmidt-apparently-bought-relativity-space-to-put-data-centers-in-orbit/

Eric Schmidt apparently bought Relativity Space to put data centers in orbit

“Gives you a sense of how big this crisis is.”…

Ars Technica
@m No one believes this will actually happen. One of the most demanding tasks of a server farm is cooling, which is notoriously hard in space. And there are no upsides (solar is out ~1/2 of the time as well, buidling is more expensive).
@tobtobxx It is frankly bonkers, isn't it? Apparently for the ISS "Outside temperatures vary between 120C and -160C (250F and -250F)", and that's just for starters. Radiation hardening will be an interesting engineering challenge too! And for all the grand visions of the folk hyping this up, the demonstrator always seems be a Raspberry Pi in a cubesat... ¯\_(ツ)_/¯

@m Well, outside temperatures don't really mean that much above the atmosphere. The solar heat could be shielded against relatively well (see JWST). But getting the waste heat away is much more difficult in a vacuum without access to cool water.

I'm uncertain about the radiation shielding, though. Those cubesats demonstrate that it be easy (as did Ingenuity), but I don't know for certain.

@tobtobxx @m fyi you can get interesting orbits which give you 24/7 sunlight e.g. Sun-synchronous orbits, + you gain about 30% by avoiding atmospheric absorption. Cooling is definitely the extremely hard part and from what I've seen nobody's proposed a good scalable solution

@Smoljaguar Oops, yeah, I forgot that. Though you do not want to go outside of of earths magnetic field.

All in all, I think this is just hype. It doesn't make sense at all, even if you require 3x as much solar arrays on earths surface, it'd still be way cheaper to build and maintain.

... it would however pour some of that investor money into the space industry, which I wouldn't object to.

@tobtobxx yeah, there are also many other issues and apparently lots of the analysis is hyper-optimistic (https://www.tumblr.com/oscillatingheatpipe/781648495436267520)
Post by @oscillatingheatpipe

💬 1  🔁 12  ❤️ 42 · There's a company out there, Starcloud, that's trying to do orbital datacenters. They claim that it solar is more plentiful, and heat is easier to manage. Seems like heat should…

Tumblr

@simon As a first and very accurate approximation, "carbon footprint" is whatever you pay for something.

The best way to reduce your environmental impact is to spend less money.

@simon this is great, but I really wish people would stop using LLMs as search engines.

@camertron complex topic: turns out "search" covers a ton of different use cases

Navigational searches (take me to the official website for company X / that article I read last week about Y) are a bad fit for LLMs

Some informational searches ("how do you get an HTML video tag to default to muted audio) are good

And then things like "what arguments do people generally make about why rent control is a good idea" (still arguwbly a "search" task) are much better for LLMs than regular search

@camertron ... and then about four weeks ago ChatGPT running o3 and o4-mini got genuinely good at using its detach tool! Five weeks ago I could have said that ChatGPT+search-tool still wasn't reliable enough to be worth using often https://simonwillison.net/2025/Apr/21/ai-assisted-search/
AI assisted search-based research actually works now

For the past two and a half years the feature I’ve most wanted from LLMs is the ability to take on search-based research tasks on my behalf. We saw the …

Simon Willison’s Weblog
@simon 🤔 Hmm, I hadn't thought of it that way, but I see what you mean. I suppose my concern mainly stems from my observation that people expect eg. ChatGPT to work exactly like Google search and treat the response as correct without checking sources. Just the other day it tried to link me to a non-existent Rust crate, for example.
@simon I think all these assessments miss the big carbon cost of devs buying bigger GPUs to play with ollama - carbon costs of making high end PCs are estimated in the 500 tonnes of CO2 range as far as I can tell, more if you go crazy

@graham_knapp @simon

I think the cheatsheet from Masley at least starts to address this, but I'd like to learn more about this area, too.