Yeah, that's probably what is most worrisome about AI-gen content is that what is produced now will be turned into future training data, once our media space becomes saturated with it. It might bring people back to appreciating authentic, imperfect human produced creativity after they get sick of the slop. But will coding become obsolete? Will other jobs be at risk? I am not sure it's doom-and-gloom yet, other than our already besieged attention-spans being plundered further. Authenticity for the news and other concerns of a "post-truth" era might be a real concern in the near future, though, while this all shakes itself out.
It might bring people back to appreciating authentic, imperfect human produced creativity after they get sick of the slop.
If you can recall how neat and creative AI art used to feel, that basically was able to be imperfect enough to allow things produced appear as things typically unseen by art as we've known it.
The problem with AI right now is that it's trying to model off of what has succeeded too well, training off of itself at this point instead of guessing. Eventually people will start blending older models with the tools of new ones, or they'll introduce some sort of noise or randomizer as a number value you can adjust, and just like that 'imperfect' will be captured and novel enough to take our place in that field.
Either that or they'll make the 'AI training' step easier for your everyman to do at an individual level, allowing you to give it a limited sample of pieces and see it stick to that look instead of the current efficiency-based mainstream.
In general I find AI art a lot more soulless than how it looked when it was still trying to figure shit out. It's now very good at doing a few things to the point of recognizing it from having seen other works like it, rather than mediocre at doing anything you want it to do to the point of barely resembling eachother. Before it was just working with shapes, now it's working with pre-expectation.
If I'm not toying around with generative Ai I'd do image to video, and feed it imagery of my own works.
I think a lot of people do this already, so there'll be less reason to roll back to older models, which are horrible by today's standards.
The last thing I made, I can't even take credit for most of it. I'm just the prompter, director and editor. The visuals were all whoever created the figurines and set. The acting was all the Ai following my instructions. The special FX, and robots were all the Ai's doing. In the end there was so little of me involved in that project, it was so insanely easy to, have made.
Chatgpt failed to make a simple mod for me for crusader kings 3 i followed all of it’s steps to fix the errors and it has no fucking clue how to code so you’re topic is bullshit