Well yeah, shouldn't rely on Wikipedia, but I would trust Wikipedia more than ChatGPT.we went from "Don't use Wikipedia as a source" to "hahaha dumb machine didn't point you to Wikipedia"
California bill (AB 412) would effectively ban open-source generative AI
- Thread starter trappedslider
- Start date
Paul Farquhar
Legend
Sure, you are comparing something that is humanly fallible to a piece of software created to make stuff up and pass it off as fact.we went from "Don't use Wikipedia as a source" to "hahaha dumb machine didn't point you to Wikipedia"
“To err is human, to really f**k things up takes a computer”.
Paul Farquhar
Legend
Good, the more flies it catches the better.
The guy in his bedroom should be spending his time creating his own ideas, not programming a computer to steel other people’s.
Better still, he should go outside and get some fresh air, whilst there is still some left.
Last edited:
MarkB
Legend
It's synthetic if it's something that has been created when prompted, as opposed to just a straight copy of a single existing work. Any new item created by such a system is synthetic by definition. They're just using the word to differentiate between generating and copying.
The important part in the sentence isn't "synthetic", it's "derived".
It would be interesting to see what happened if somebody pirated a bunch of Meta stuff and they took them to court. I mean, that wouldn't affect the court case, but how it played across the media.
trappedslider
Legend
Actually the little guy is whoever can afford a decent GPU,if i had a knowledge/want I could train a limited model at least for stable diffusion on my rig using a Nivida 3060. It's not going to be a large one but i could do it.
Civitai: The Home of Open-Source Generative AI has hundreds of models all made by various users.
Not a Decepticon
Hero
Hope this bill passes and is adopted in other states and countries, I wish all generative AI companies bankupcy.
TBH those are mostly (rather small-scale) modification to base models. The cost of training is however lowering for base models: that was the reason the Chinese text model Deepseek got a lot of attention. Not because it was better than the existing top players, but because the compute time cost was only 6 millions USD instead of the (presumably) billions of OpenAI's.
A model image would need less training. AuraFlow is the leading model for prompt adherence and was developped by a little guy (student at the time) (although helped by a small compute grant). Pony v7 cost 300,000 USD of compute time and is a community-funded initiative paid by little guys. And it's an ambitious base models. The cost might also be declining in the long term because one doesn't need to start from scratch by using a decent base model, though. As you pointed out, a good base model might only need limited finetune to be improved, for which a gaming computer is enough.
Also, a H200 right now starts at 100,000 USD. People routinely pay that price for cars. They are not "little" guys but the type of system needed went down from extremey costly to within the means of an enthusiast. And it won't become more expensive with time.
Last edited: