I think in internet culture, memes, and modern day art, there've been almost clear cut cycles of ideas and popular tunes of which ideas web in and out of popularity. In the earl 2000s, memes primarily consisted of images and text, of which the user likely just cropped and placed. In the latter half of the 2000s, and earls 10s we saw a huge rise in popularity of scrappy, messy, almost bad, memes a la rage comics. This then reverted somewhere over the next 5 years, back into more crop and paste text styles a la snapchat and tiktok. Popular media on tiktok started out with dancing behind text, or primarily text focused videos, but post-pandemic we saw a shift to satirical, rough, and messy video formats.
And in the same way, as of May 2nd, 2025, all AI art/media generation is characterized by how perfect it is. "Oh Sora can generate fingers with the correct number of fingers". "Yeah but Runway understands object permeance". We're currently striving towards picture perfect content generation. Any imperfection is called out, and we take our attention away from content that has a trace of error in it. But this won't last forever. There will come a day, when I ask Sora to generate "a slow motion video of an 8 armed humanoid beast " and it will provide me with 4k, picture perfect, coherent video. Gemini will generate me that perfect example of what a watermelon would look with with a strawberry-like interior.
And we'll be happy with this. And rightfully so!
But, like all things, this too will come to be unfavored. Soon enough, we'll be craving for the chaos, just hoping our token request comes back with a smeared fingernail, or an extra arm. Anything to hide it's perfected ability.