Edit: In other words...
1. Train your stable diffusion model using a corpus of normal human images.
2. Use the model to generate thousands of images.
3. Use human analysts to train another AI how to find "good" images in that set.
4. Generate ten million stable diffusion images.
5. Use the other AI to find the "good" ones.
6. Use those to train another stable diffusion model.
7. Goto 4.
(You lose some information each iteration, though, so it would probably eventually converge to grey goo or millions of pictures of Agent Smith.)
https://paperswithcode.com/dataset/simulacra-aesthetic-capti...
https://github.com/LAION-AI/laion-datasets/blob/main/laion-a...
https://en.wikipedia.org/wiki/Generative_adversarial_network
That's their analogy to ML/AI art: imagine it takes over and puts most artists out of business... are the models just stuck, taking their own output as input and not really improving? Similarly here it's maybe not possible for this to happen to such an extreme degree that the imagined problem would come to pass.
It's a common fallacy that things need to work for everyone or they're no good. We can do different things. Specialization is useful.
-
At what point will we doubt reality and truth in favor of the AI provisioned one?
We talk about singularity being an AI that surpasses human int... but its goig to happen much sooner and via a diff vector, such as AI images.
the fucking speech by biden the other day looked like a deep fak to me - his mouth flapping was way exaggerated, and the fact that he could hold a salient sentence were the tells...
singularity isnt about AI being able to fool us that hte __*AI*__ is a human - its that we FOOL OURSELVES...
So I think that the whole AI art thing is going to make reality so elastic in our minds that it will be likea dream, and we litereally cannot trust a single fucking thing online. period.
AI is going to kill the HUMANITY. (this is a phrase that one should all know... HUMANITY)
AI is killing it. and gues who is trying to own AI - fucking pedo-financial-hedge-funds
If you read it too much, and go back to normal human forums you will wonder if they're also generated by AI or if we are all just AI generated bots on the internet :-O
I'm not sure what you mean here. Are you saying that index funds are driving individual equities? If so, why do you think that? Because of broad market correlation?
It’s a great metaphor for what could happen if there’s no new art coming from outside the model; everything get stuck (though I’m not sure I agree, tbh, humans are just pattern-matching robots too and it seems to have worked ok)
The algorithmic art is similar in the sense that it is an averaging of previous effort. People made art and it was fed into the algorithm. The algorithm is some weighting of the art it was fed. Analogously, people who thought about companies bought and sold the stock. It landed at some price. The index fund just followed that.
There has been criticism of index funds [1] and defense [2] as far as price discovery.
Other people have raised the possibility that an index fund holding large amounts of competitor companies is an antitrust issue. [3]
[1] https://www.ft.com/content/87a3b117-7085-40f4-8cca-66bb8353c... [2] https://www.blackrock.com/corporate/literature/whitepaper/po... [3] https://corpgov.law.harvard.edu/2018/12/04/why-common-owners...
It's a little like copying the work of hundreds of other artists at once.
Index funds bother me in the same way as in at some point the price discovery mechanism ought to break as everyone just uses one corpus.
All in all, this is an amazing idea, enabled by the work of amazing people, and it is like looking at something significant - I just don't know what :-)