I think this is going to hit like in other industries like programming, and disproportionately affect new artists, artists that are themselves still learning what they like.
Some “tech forward” artists will try to not fight the wave, start using AI, and their drawing skills will never develop, leaving them dependent on it with a ceiling to what they can produce.
Other artists will be blocked and they can never jump from the high-school doodle to one-shot to series steps because the quality curve will become a 90° wall.
Other artists like Inio Asano or similarly innovative newcomers who are just legitimate geniuses will break through, because AI can’t come close to having so innovative or compelling authorial or artistic voice.
“I am now competing with illustrations that may have been trained on my own work without permission.”
That’s fucked up.
That’s roughly similar to catch up development (which is always faster and easier), except automated. With catch up development those doing it only make waves for some time before they, well, catch up and find deep inability to evolve, similar to USSR, which is where it ends. While with this - things are made from something already created all the time.
Basically all you need to make something a source of training data, laundered of authorship with plausible deniability, is for that something to be publicly available.
Meaning that at least in art we might be back to private orders and works available to limited circles of viewers, I think. Once it becomes unprofitable enough to publish your works, and once the existing pool of training data is exhausted and falls back enough compared to what’s in demand and in fashion.
OK, that’s just an exercise in high school level “outta my ass” style analysis.
I mean, Antique Mediterranean had all sorts of mystery religions. Perhaps we are on the threshold of an era of mystery art.
There’s already mystery music out there. Small circles of folks putting things out that they don’t expect or want people outside the circle to hear. I’m a nobody and I’m part of two different groups who share music with each other, build on each other’s works, try genre mashups and new shit that may never get done again, many times because it’s a mess but sometimes just because it was a fun one time thing.
That’s really interesting. I’ve mostly done only small experiments with generative music and povray renders, sharing them with friends, but that’s usually not enough to call a circle. But my friend has such a circle for poetry. I mean, with poetry this is pretty traditional.
“may have been”
RIP Anime?
Is the future just gonna be filled with AI Slop?
Probably kinda.
I imagine that someone who can’t draw for crap using AI to create the visual component to their stories. They may also use AI to create character “voices” who say things in their specific manner
Manga production costs drop as a writer doesn’t need as many skills to write them and the AI handles more of the heavy lifting when it comes to generating content.
Eventually one of these slop mangas becomes popular enough to get animated. Since it is animated, it becomes a lot easier to make. My guess is that the beginning looks like AI empowered rotoscoping, human actors provide the movements which become the basis of animated characters. You can even do voice modulation too, so the vocal performance gets swapped to the right voice.
These animes might be less popular, but if they are being made at a fraction of the cost of human animation.
I think short term, yeah, I think so. Medium term and I think we’ll see a bunch of model collapse or people will get tired of the same story line repeatedly so it won’t be profitable (or won’t be regaining the money they thought they’d save on actual creative folks). Long term, I have no fucking clue if they’ll get around the fact that training AI on AI makes it even weirder.
Yup. Mountains of slop is gonna replace schlock. What was before schlock will soon be slop. Well, it won’t be the end of schlock entirely. I’m sure we’re still gonna get some schlock that’s also slop. Slop schlock.
I’m surprised it is that low to be honest.
The overwhelming respondent percentages are impressive once you know that nearly 25,000 people were surveyed.
