In our fervor to prioritize humans over machines when evaluating art, we’ve conditioned our minds to hate anything perceived as “AI/slop” and love anything perceived as non-AI. In visual art, AI “slop” typically manifests as eye-popping saturated colors, heavily made-up, overly beautiful people, and post-processing. In short, AI was trained to produce what people considered beautiful, and now, in an ultimate uno reverse card, everything that people considered “beautiful” is now considered ugly, because it’s associated with “AI”. This effect is happening across all mediums: Visual art, music, and writing.
Case in point #1: Observe how in this Reddit post, everyone is gushing about a new AI model, noting Image #2’s superiority over Image #1. What everyone seems to be ignoring is that the main difference is simply that Image 1 is what a professional photographer might’ve done after working on it for hours and applying lots of post-processing, whereas Image 2 is what a normal person would’ve done just be snapping a photo on their phone. If you’d asked everyone 10 years ago which image is better, everyone would’ve answered Image #1. But today, everyone answers Image #2 because it looks plain, which means it’s not as “AI”.
Case in point #2: In 2023, a news article circulated about an artist who was banned from an online community because his art closely resembled AI art. They told him to “find another style”. Never mind that his works are part of the reason AI could make that type of artwork in the first place, and the reason AI was trained to produce that kind of style, was because it’s what people considered conventionally beautiful at the time!
Case in point #3: Many bots posting LLM-generated text, like the one in this reddit comment, are now programmed to purposely make basic grammar mistakes, such as avoiding capitalizing the start of a sentence, and to replace the EM dash with a hyphen. In short, people posting AI-generated text are compelled to literally insert mistakes to make it seem more attractive and less like AI.
At this rate, in the future:
- “Good” visual art will require color-correcting your image to look as bland as possible
- “Good” music will require making so many music theory mistakes that an AI couldn’t have possibly come up with it
- “Good” writing will require lots of grammar/spelling mistakes and misusing hyphens as EM dashes.
