What the poem Google made me can tell us about art
Picture the scene: I was working on a naming project where poetic form names seemed like a viable theme—Sonnet and Haiku and Villanelle and the like. There was an obscure form I remembered learning about in my undergraduate poetry minor, but I couldn't remember the name of that form, and I didn't see it referenced in any lists of types of poem. So I googled its main attribute: "seven stanza poem". At the top of the list of results, Google's LLM decided to just produce me a poem with seven stanzas, on the subject of the city I live in. The poem—and I feel I hardly need to tell you this—was awful.
This experience was distressing, I thought, for two broad reasons: technological and artistic.
The tech reasons, you already know. They're the same problems we've had with tech companies for the last ten years, with LLMs for the last few, and that I’ve written about quite recently. Let’s dispense with them quickly:
Absolutely unrelated to what I was actually looking for, because tech companies now rarely care about why their users use their products
Based on a literally unimaginable user story of a guy who just wants to read a poem, any poem, and doesn't care about anything at all except its structural length
Full of AI hallucinations: for instance, the LLM identified the poem as "free verse" but was entirely in (roughly) AABB pentameter.
None of the search results were what I was looking for, either!! (which was the word Sestina, by the way)
Privacy nightmare, why does Google's LLM get to just know my location
Optionality nightmare, as there is no way to disable these technologies
The artistic questions, though, seemed more unique: the one thing I WOULD sort of expect from a computer-generated poem would be precise technical perfection, but in addition to being bland and purposeless, this poem was also just clumsily constructed, with poor rhymes and awkward meter. My brain, which has the special blend of analytical and creative instincts that made me highlight the various extended metaphors in my workshop cohort's poems and scan every line's meter, immediately kicked into critical overdrive, and I began a serious technical and creative analysis of the poem.
But the more I thought about it, the more I realized that there wasn't ACTUALLY an artistic critique to be made. For me, one necessary factor1 for considering something “art”, is that it has to be a product of creativity and intention. A painting of a sunset can be art, but the sunset itself is cannot2. It may be beautiful and pleasing to see, or it may be washed out and disappointing, but those differences are only artistic if you're talking to God.
This is what makes art uniquely interesting, and talking about art interesting. Why did Robert Ryman paint so many canvases white? After all, he did choose to do that, and must've done so for reasons, blending aesthetics and politics and technical skill and surely countless other conscious and subconscious factors. Speculating about the myriad possible answers to this question, or doing research to learn more about them, can be incredibly rewarding, which is why art criticism and art history have been such enduring and popular practices.
And this question—why did the artist do this, in this way?—sparks so many other interesting questions about our experiences and our society. Why would a museum curator, a very different kind of person than an artist, choose to acquire a piece like that and hang it up? What does it make me, looking at it in this museum, feel? Why does it make me feel that way? Do the people around me look like they feel the same way about it? Why might that be? What knowledge or belief or perspective or value might make them think differently? Why was someone willing to pay millions of dollars to own one of these white canvases?
These are the kinds of questions that transform a work of art into a window into human souls. They are what I find most interesting3 about art. That you can ask these questions about a canvas painted white, or the 1993 film Tombstone, or Sarah J. Maas’s A Court of Thorns and Roses, or your nephew’s crayon scribbles, seems to me like a signal of what we mean by the word "Human."
Most of these questions can't be meaningfully asked about, say, a sheet of printer paper. It, too, is an artificially whitened rectangle, and it is a product of human labor, but the answers are all obvious and exactly the same: "it was mass-produced to be sold to people who wanted to put pigment printed onto it." There are other questions to be asked from an anthropological view, or economic, or technical—why did 21st century humans need to reproduce so many words, and what made this the ideal form factor?—but the artistic questions are pretty open and shut.
These questions absolutely cannot be asked about an overcast day—another big white square, but one where the answer to "Why not another color?" is answered unambiguously by physics.
LLM-generated content, especially this sort of unsolicited and entirely whole content, has more in common with a sheet of paper, and even a cloud, than it does with a Ryman canvas. It does, technically, lie at the end of a series of decisions made by people, and many of those people clearly care quite a bit about aesthetics and politics and technical skill. Yet the output itself is entirely divorced from any of that: to ask "why was it so clumsy with the pentameter" or "why did it number the stanzas" might have some esoteric interest to LLM programmers, but it has nothing to do with souls of any kind.
So, what does the poem that Google made me tell us about art? Nothing. Although I could not prevent my mind from exhaustively critiquing of Google's work, I will not waste your time with that critique. If you want to read the poem yourself and ruminate on its merits, I won’t stop you.
But I hope I’ve already dissuaded you.
If you'd like to read some more human-generated content, consider liking this post, subscribing to the newsletter4, or even contributing your own content by commenting below. You might also want to follow my new project on Bluesky, called Apropisms, where I will tell you about exciting new idioms that you should definitely use.
See ya next week,
Though not the only criterion. While every artist seems to have their own definition, human creativity and action is a common theme.
Unless chemtrails are real and you are doing them specifically to change the color of the sunset, in which case, honestly go off king.
Of course, "interesting" is only one of the qualities I'm looking for in art. Quite often, "enjoyable" is much more important.