This is something that began long before AI art. Look at the popularity of Thomas Kinkade - The Master of Light! Many AI generated images, when not refined down, now have that same sort of aesthetic. I have found that Googles Nano Banana generation is particularly susceptible to generating that cozy glow kind of style than many gravitate to.

It is art stripped of the humanities. Pleasant, non threatening, technically very capable and totally disposable. The Big Mac of creativity.

The author’s primary claim seems to be that AI art defaults to what most people prefer (‘accessible’, pleasing to the eye/ear, etc.), which he puts in the bad category even if created by a human. However, I did not notice whether he checked if AI can generate images/poems that aren't so accessible/aesthetically pleasing, by asking AI to do so? It seems like a huge part of his claims about AI generating art (that's similar to bad human art) assumes that accessible/aesthetically pleasing art is all AI can do, and it can’t do otherwise if instructed so.

Otherwise, I guess the argument would devolve to: AI defaults to generating art that humans prefer (‘bad’), but AI can also do far more as well (if asked) when generating images/poems? (Although OP article is from '24, so capabilities may have changed.)

I think there’s been this huge transition over the past 50 years from a sort of systemic-humanistic view of society and a purely sort of maximalist-efficient view.

I think some people still have this view but like art is a process created by institutions and tastemakers in places like New York City and Hollywood. The fact that there’s a narratives that a creation exists in a node of a graph of your complex model of existence yoU hold in your head in time, in space, An artifact with a story.

And this transitions to this idea that the best art is the one that combines pixels on a screen in the formation that’s most stimulating certain nerves in the back of your eyeballs.

It’s a profound change.