this post was submitted on 02 Oct 2024
366 points (100.0% liked)
Technology
70415 readers
3407 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- 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, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I feel like the last few months have been an inflection point, at least for me. Qwen 2.5, and the new Command-R, really make a 24GB GPU feel "dumb, but smart," useful enough so I pretty much always keep Qwen 32B loaded on the desktop for its sheer utility.
It's still in the realm of enthusiast hardware (aka a used 3090), but hopefully that's about to be shaken up with bitnet and some stuff from AMD/Intel.
Altman is literally a vampire though, and thankfully I think he's going to burn OpenAI to the ground.
What do you think about the possibility of decentralized AI through blockchain so that you could pay some tokens or something like that to rent the GPUs to run your AI for as long as you wish to instead of having to buy all the hardware and assemble it yourself?
I cannot tell if you are being serious or just having fun with buzzwords
I stopped at block chain
Oh, I'm being completely serious. I've been interested in crypto since about 2013.
crypto fizzled years ago without a major use case. source: check google trend history.
bitcoin wouldn't be useful for tracking the rental of cpu/gpu assets.
Yeah, no Bitcoin would be the wrong chain for that. But there are other chains that would work better for such a use case.
how do you suppose?
Bitcoin is really too slow and too expensive for automated tasks like that. There are other chains such as Solana, etc that are much faster and much cheaper that would work better for that kind of use case.
There is no need to use Blockchain for this. Computing pools like this have been used for ages
The solution to all the world's computing issues is not "blockchain."
you mean a computing pool, like SETI@home since the late 90s?
absolutely no need to make this idea stink of a crypto scam.
You can already do it but there isn't really any need for a blockchain. I personally use runpod but there's vast.ai and a few others.
It's usually quite cheap.
Isn't that just cloud computing but with extra steps?
Tbh it's just hard to see the value proposition in the age of cloud computing. I think aspects of the underlying technology are cool but basically every crypto project that comes to mind has been an actual scam. Sure there's eth and RDNR that was built on top of it but why should i spend what will ultimately be more money in periods of high demand (gas goes up when more people use the network) when i can just plug my credit card into amazon or microsoft AND get the benefit of infosec regulation like PCI-DSS. Crypto just doesn't ever inspire confidence because bad actors consistently shit in the punch bowl while providing no extra utility over existing cloud providers.
When distilled down crypto-compute just seems like cloud compute with extra steps, which is already just using a computer with extra steps.
We already can rent GPUs to run AIs with tokens - those tokens are just managed by govt instead of some random.
I somehow didnt' get a notification for its post, but thats a terrible idea lol.
We already have AI horde, and it has nothing to do with blockchain. We also have APIs and GPU services... that have nothing to do with blockchain, and have no need for blockchain.
Someone apparently already tried the scheme you are describing, and absolutely no one in the wider AI community uses it.