Some people feel the need to attach an image to almost everything the post - even in chat rooms (including slack at work). These images serve no purpose, but the poster feels it is useful - though it was often reaction gifs in the past I'm seeing a lot more AI content than I ever saw reaction gifs, maybe because of novelty or maybe because it's ultra personalised.
Sometimes images help to visualise something or to get a point across, but I see so many people who think it's necessary to reply to a discord message with a cat with human limbs doing a dance, or a photo of "themselves" climbing a mountain with the Rust logo to show them mastering Rust... Ok?
Image models are useful and I'm thankful for much better visual reasoning but it really really frustrates me the constant need to burn money for all of this slop.
>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
If anything now I feel LESS guilty about usage.
* if we trust the numbers in the other comment
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.)
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.
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.
(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.)