The link between art and life

Nabeel on AI and art

I wrote about Jonathan Swift’s Frustrated Humor ahead of the anniversary of one of English Literature’s most important books, for Plough (it will be appearing here in a couple of weeks)

---

And yet, though I’ve pointed out many features of great art, there is still some mystery to it. This is, I think, the link between art and life. There’s a certain weight to someone having had an experience — or having imagined something — and then having written that down in this way. It’s the type of weight you feel when a grandparent tells you a story that’s important to them, or when someone shares something especially vulnerable that happened to them.¹⁰ This is, ultimately, why I think human art will continue and even improve in the age of AI: AI cannot experience things for you. Art, at its best, is in part about these most spiritually weighty human experiences, and this component of greatness is not reducible to any formal factors, because it is life itself.

Nabeel S. Qureshi has written about “what makes art great” and why AI art is (currently) not great. I don’t disagree with what he says, though I take a different view of how surprise and patterns work, and I suspect Nabeel would too if he was writing at greater length. I want to add a brief thought about the way surprise and pattern interact, and to speculate about AI writing based on that interaction. My thoughts here are a bit loose… because I don’t think this is the sort of topic on which it is possible to be very confident.

If art is “not predictable or obvious” and “contains multiple overlapping layers of echoes” then we discover a conflict. Those echoes are not only internal to the text, they exist between texts. That is part of how a tradition is created. In simple terms, red, white, and blue flowers in Chaucer mean it is spring and someone is going to fall in love because that is what they mean in other poems of that nature. It is a trope.

Art has to conform to these tropes and traditions and work to be original. For many authors, it has been more important to work in imitation than to “discover your own voice”—and for many modern authors who think they are developing their “voice”, it is more true to say they are conforming to some sort of genre or sub-genre. This is just as true in literary fiction as in other sorts of fiction.

Nabeel gives the example of the repetition of ears in Hamlet. Obviously, Hamlet is a play of spies. Claudius says “When sorrows come, they come not single spies / But in battalions”; Polonius hides behind the arras; Hamlet is deceiving multiple people; and so on. The court is full of noises, gossip, and strange rumours. This was not an uncommon trope at the time—whispering poison in someone’s ear was a familiar concept at the Elizabethan court, no doubt, as was the idea of being overheard by spies, personal or governmental. It is common for characters to eavesdrop on the Shakespearean stage. Maybe the audience would have thought of Matthew 13 and the parable of the sower, when Jesus says, “He who has ears to hear, let him hear!”, a trope found earlier in the Bible.

The “ears” in Hamlet work somewhere between a predictable pattern and a surprising one. It is well known that Shakespeare is the most surprising of writers—he is much less predictable word by word, line by line, than other writers. But he is often surprising with predictable images and tales, as with the ears in Hamlet. There were many versions of Romeo and Juliet. We remember Shakespeare’s. It was surprisingly familiar, and still is. Joyce is remarkably surprising, often about familiar things. This, I think, is Nabeel’s “link between art and life”. Martin Amis liked to say, “it’s life, but is it art?”, when in fact great art usually has to be both in some manner.

All art has some moral purpose—it surprises or familiarises in order to call us to the things of this world. It is that calling, that link with life, that leaves some people to conclude we will never have great AI writing. It has to be written by a person to be interesting. But some of what poetry does is the arrangement of sounds—what Elizabeth Bishop called timing.

In physical life, Bishop explained in an essay about Hopkins, timing means “co-ordination: the correct manipulation of the time, the little duration each phase of the action must take in order that the whole may be perfect.” The time of each of these parts of the action is decided by “the time of the whole, and of the parts before and after.” Bishop gives the example of men rowing a boat, and all the tiny motions going into each stroke, and how they amount to a rhythm. That rhythm, she says, is what is happening in poetry.

Just so in poetry: the syllables, the words, in their actual duration and their duration according to sense-value, set up among themselves a rhythm, which continues to flow over them. And if we find all these things harmonious, if they amalgamate in some strange manner, then the timing has been right. This does not mean that a monotonous, regularly beating meter means good timing—duration of sense and sound each play a part, I believe, nearly equal, and sense is the quality which permits mechanical irregularities while preserving the unique feeling of timeliness in the poem.

We might think of this amalgamation as the essential happening of poetry not just at the level of timing and rhythm but also of meaning and moral intent. It must all hang together. As Bishop says, sense is what permits irregularities in sound. It is this amalgamation that is required to make the link with life live in the poem.

It may be, as some have argued, that AI simply cannot achieve this, or at least, cannot achieve it in any way some readers wish to experience. But we are in the middle of an experiment. gwern’s poems co-written with LLMs are good enough that, had they been presented in a blind test, they may well have passed.¹ I still suspect that AI, left to its own devices, will have its own taste. But the models are getting better. Prompters like Gwern are well-attuned to when a poem is properly amalgamated. Perhaps we are about to see a new type of poet and we will have to change our minds about what it means when we experience great art.

1. Read Hollis Robbins on this topic too…