Comment by paimapi
18 hours ago
plus the energy cost - https://www.technologyreview.com/2023/12/01/1084189/making-a...
>Generating 1,000 images with a powerful AI model, such as Stable Diffusion XL, is responsible for roughly as much carbon dioxide as driving the equivalent of 4.1 miles in an average gasoline-powered car. In contrast, the least carbon-intensive text generation model they examined was responsible for as much CO2 as driving 0.0006 miles in a similar vehicle
That's the complaint? That if I make pictures for a few years, let's say on average 1 a day.. it will equal a small errand of carbon?
If anything now I feel LESS guilty about usage.
But 3 billion pictures per week is equivalent to 12.4 million miles or 639.6 million miles extra per year.* It's not about you. There are no 3 billion people each creating one picture per week. Its probably more like 10.000 assholes creating 2.5 billion pictures per week to satisfy some stupid online feed and the other users creating a picture per month on average.
* if we trust the numbers in the other comment
I mean that in a big number..
It equals around 0.004% of actual driven miles if my estimates are right.
That is with an estimated ~270 billion miles vehicle miles every week (I put commercial in there.. about ~200 billion miles if you only include passenger vehicles.)
That study is from 2023 when the cost to produce 1000 images was 2.91 Wh per image. More recent numbers for Stable Diffusion's 2025 models that puts it at 1.3 Wh per image and that number continues to decrease.
For context that means a days worth of image generation emits about the same amount of carbon dioxide as a single transatlantic flight (New York to London). There are approximately 1500 transatlantic flights per day.
So every day the people of Vermont emit more CO2 just commuting than all image generation through OpenAI per week. Nice. That does put it in perspective. It’s really efficient.
yep, what our climate really needs is another carbon emitter similar to an entire state's worth of vehicles just so people can not pay artists or make dumb images of themselves ripping off some artistic style like Studio Ghibli. that is so much more of a social good than people being able to make it to work and earn a living
Oh I think paying people to make 3 billion images would result in far more carbon emissions. Just think about how much CO2 a human emits while painting. The paint, the food. Just for the sake of the climate I would never pay an artist for something like this.
2 replies →
1000 images is worth 4miles?
Ours cars are atrocious. We can have the most powerful artificial minds imagine 1000 images from simple prompts, and that takes as much energy as moving a human being 4 miles.
And that will get more energy efficient. Gas cars have barely budged.
Gas cars are the problem, not AI.
This is an embarrassingly idiotic article. They're comparing against the least intensive text model, ie some million parameter model nobody uses. Stable Diffusion is a 3.5 billion parameter model, while the GPT models a billion people are using for text generation are over 10 trillion parameters. To say nothing of the differences in average context size usage. The actual ratio of SDXL to text generation pollution is probably literally reversed from what the article claims by lying with statistics.
(Note, however, that OpenAI's image model is much much larger than SDXL; however, we don't have precise numbers for it. Nonetheless, misinformation is misinformation.)