AI portraits don’t even look real so AI porn is going to be heavily uncanny valley. You’ll probably have a penis that looks like a hydra head and seven testicles.
Nope. This is all due to the amount of training done on the material so far (very little).
If you actually follow the research papers, computational power and training hours done, you can quite intuitively plot this on the graph.
I'd give it max 2 years that the videos are good enough for the average user to be good enough. Max 3 years and the state of the art will be so good that most don't care.
The rest is down to economics: how much does it cost to pay real actors a cut, how fast does computation /training cost go down and where's the cut off point. That one is much harder to guess, but most likely will happen within 10 years (could be much sooner).
This is the future : all digi-ID strong biometric auth to use anything and most porn will be AI generated, making droves of 20-somethings unemployed.
The only question : who benefits?
The rest is noise and there's nothing any of use can do to stop it.
Just Disengage and go back to real life.
AI portraits don’t even look real so AI porn is going to be heavily uncanny valley. You’ll probably have a penis that looks like a hydra head and seven testicles.
That's my fetish
No. The best AI Portraits have been indistinguishable from real ones for years already, unless you are a trained photographer/cgi expert.
https://thispersondoesnotexist.com/
The finger thing? That's not part of a portrait normally, but that's been solved for still images now.
You just need enough training.
Video is still much harder, but again, like I wrote: a function of training hours (material) and computational power. Rest is economics.
Only works for the face. AI once you go past the neck looks like some kind of monstrosity.
Nope. This is all due to the amount of training done on the material so far (very little).
If you actually follow the research papers, computational power and training hours done, you can quite intuitively plot this on the graph.
I'd give it max 2 years that the videos are good enough for the average user to be good enough. Max 3 years and the state of the art will be so good that most don't care.
The rest is down to economics: how much does it cost to pay real actors a cut, how fast does computation /training cost go down and where's the cut off point. That one is much harder to guess, but most likely will happen within 10 years (could be much sooner).
https://www.youtube.com/watch?v=MikbhfH28A