this post was submitted on 22 Nov 2024
53 points (75.7% liked)
Technology
59569 readers
3825 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
The whole planet is threatened by AI. If you look at the amount of energy needed not only to power the infrastructure, but the energy needed to create the infrastructure, and compare it to the work produced, and the energy needed for humans to produce equivalent work, it's totally fucked and dumb as hell
Edit: to elaborate, there was this commercial for a Google Pixel I saw, people in group chat talking about a football game, person says "create an image of football gloves made out of butter". .08kWh later that image gets posted in chat for a chuckle. Dude, just say "gloves made of butter? smdh" Lady laying in bed talking to a glorified chatbot, just hop on Lemmy or reconnect with an old friend and save .16kWh. These are the most common use cases for AI, basically finessing a prompt a dozen times to make a Shrek and Garlfield comic that winds up in some Facebook group with 9 likes. Multiply those figures a couple million times tho and it's like holy shit. We somehow went from extremely low-bandwidth words to high bandwidth Youtube and Tiktok to the messiest bullshit humans have ever invented, to do things we could easily do with characters on a keyboard
Energy demands are only going to increase as we replace gas with electric alternatives. The problem you're pointing to is an issue with the current infrastructure.
Infrastructure in this case refers to the data centers and LLMs. It takes hundreds of megawatt hours to train a single current-gen LLM and who knows how many gigawatts of energy are being consumed by the sum of LLMs at any given point but it likely dwarfs the sum of all energy spent training LLMs.
But then there's the energy involved in producing those cards, shipping those cards, the data centers themselves.
It wouldn't be preposterous to suggest that the sum of energy spent at any given time on generative AI is enough to power New York City. Might even be well more than
No, in this case I'm referring to the electric grid and what powers it.