this post was submitted on 16 Feb 2024
167 points (94.7% liked)
Technology
60022 readers
3338 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 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
For now, but with every new tech, hardware efficiency optimization is not too far behind. Especially considering the performance required for training != performance required for running/outputting.
Considering the glacier speeds our government moves, I'd bet on those hardware efficiency optimizations making it out before any significant law gets implemented.
Wait for ASICs. When I started mining bitcoin on dual core cpus I would have never believed a little USB stick would be doing 100x what 30 of those full sized pcs could. The software behind generative models still needs to be fleshed out for hardware to know what direction to go. But a little AI machine that only draws like 50w max but has 256gb of high speed RAM for model storage is not far away. Anyone with $500(or whatever idk) will be able to order a machine off ebay, load it up with whatever models they want, and start generating whatever they want locally.
I don't think it'll be as easy as calculating SHA256 hashes, so ASICs as small as this might never be a thing.
On the other hand, brains do use orders of magnitude less power, so who knows.