Aside from being both unethical and unsustainable, AI images are just plain ugly. 

Recently, we’ve seen the rise of AI image generation, along with all its negative consequences. Companies are firing artists in favor of AI, misinformation is being spread using AI and the natural resources it takes to maintain AI servers are incredibly high.

However, some people still generate AI images, even thinking that it’s better than actual art. Oftentimes, they don’t understand how AI image generation works, and most times, they don’t even realize they still prefer human artists.

Stable diffusion models, which most AI image generators use, work by starting with random noise before refining the image to look more like the prompt it was given. With many more iterations, the image starts to look more and more like the target prompt.

Unlike the idea that AI somehow blends images together, this process of iteration seems similar to actual artistry. However, the differences are what make this process so ugly.

First of all, there is no tactile form of creation here. There is no drawing, no photography, not even printing. This removes the human aspect from the art entirely, and as many people (artist or not) will tell you, the process is part of what makes art special.

Secondly, as close as it can get, stable diffusion models can never and will never be able to create an image without mistakes. After all, it isn’t creating at all — it’s imitating other people’s creations.

AI image generation is based on data sets that incorporate other people’s art, often without their consent. Using this data, it copies patterns and traits in order to generate an accurate image. 

However, it cannot make something that it doesn’t have any data for. Since it is just recreating patterns that it recognizes in its data, it cannot create its own patterns.

Furthermore, AI lacks one very important part that forms the basis of all art: imagination.

When prompted to make an image of a person riding a horse, AI will easily be able to. However, when prompted to make an image of a horse riding a person — something most people can easily imagine — AI will fail. 

Since the keywords are “horse riding,” it will imitate the data that most closely resembles those keywords, which show a person riding a horse. However, it cannot break this pattern at all, and therefore has limited ability to create.

Art can infinitely evolve. Each piece is inspired by the ones that came before it. Each person brings their own style to art that no one else has. The same cannot be said for AI.

The more you look at art, the more you gain. You see details you hadn’t noticed before, or signatures of the artist, or even mistakes that you yourself would make. The more you look at AI images, however, the more you lose.

Some people believe that art is measured by how beautiful it is, and that it can be measured. The beauty of art, though, comes from the time and care put into it. AI, which can spit out images in seconds, can never compare.

No parent has thrown away their child’s drawing in favor of an AI-generated image. After all, no matter how well it can imitate others, there is no replacement for the artist themselves.