I just want to point out that the insane "several steel foundries" worth of energy waste are used in training the bots, and individual requests are, by comparison, much cheaper.
Mind you AI "art" is bullshit and AI "artists" are proof that more people need to be shoved into lockers, but we don't need misinformation to hate AI and AI bros. It's plenty hatable with just the truth.
100% this and the narrative that OP is promoting serves to shift the blame onto the consumer, when in fact AI-companies are not profitable and only works because investors and the US government are willing to funnel billions into them.
There’s bots pushing this narrative en masse because the tech billionaires are lobbying for monopolizing control and access of AI. OP is either a bot, or was fooled into buying the propaganda.
I've heard a lot of compute is starting to be shifted to inference nowadays with all these chain-of-thought models
Interestingly enough, you are correct, and I read about it in the same article that is being depicted in this comic. (MIT's paper on the subject.)
https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117
I'll tell the disabled kids I work with that you think they should be shoved into lockers because they like playing with AI art.
Yes, yes, everyone: hate the AI, focus on the AI, think only of the AI... We'll all be put on special trains for meatbag surplus soon enough. Now, back to the AI, citizen. Haaate. Focus. Run on your wheel.
You left out the part where the user has to generate multiple images because the initial ones are absolute trash.
That implies that following ones... aren't. 😶
implying they go to that much effort
I think using ai for some stuff isn't that bad, like explaining complex stuff or as an alternative to a search engine, but if people wanna use it they shouldnt use that ultra power intensive stuff that those ai companies offer, but rather one of those locally hosted models on their own machines.
To me that's so cringe because I've tried it out for explaining concepts and when I take that information and try to use it, it is confidently wrong so much of the time.
The one thing it has helped me with is when I'm trying to do some system administration task, where traditional search engine results are old forum entries or out of date documentation, llms can suggest a way to do the task, and then I can follow those breadcrumbs and do real research on how to do what I need to do.
I was trying to think of use cases for it. honestly if you just want a general overview of a topic, the hallucinations dont affect it too much
Wait till you find out about crypto and how much energy it uses compared to AI, now THEN you will lose your shit
Reminder to everyone that a lot of this anti-AI propaganda is being pushed by capitalists who want to cut off access to AI output for everyone but them. It’s part of their plan for entrenchment of technocracy.
