Gloomy Vanity

 Everyone knows that technological change sometimes displaces whole art forms. It is a commonplace of art history that the invention of photography totally reshaped the purpose of painting, because oils and pigments were suddenly no longer the most efficient way to accurately represent a visual scene. Painters thus went off to explore more subjective and emotional impressions that were not so grounded in the mere literal transcription of reality—hence movements like expressionism, impressionism, Symbolism, cubism, surrealism, etc. 

In his introductory essay in his collection of prose pieces, Aren't You Rather Young to Be Writing Your Memoirs? (1973), B.S. Johnson points out that something similar happened to the novel after the development of film. He notes that Joyce was well aware of the potential of the cinema to be a more efficient medium for telling stories. And so, the novel had to figure out what it could do better than film—if its 19th century task of merely telling "what happens next" in a narrative had been eclipsed. 

And so, the novel—like painting—had to become more subjective and introspective. As Johnson points out: film and television might be able to depict perfectly well "what happens next"—but they are less well equipped to describe a character's thoughts and subjective consciousness. The novel of interiority was born. The critical concept of "stream of consciousness" was coined (by May Sinclair, specifically); etc.

As Johnson puts it: "the novel may not only survive but evolve to greater achievements" in the age of film and TV "by concentrating on those things it can still do best: the precise use of language, exploitation of the technological fact of the book, the explication of thought. [... T]he only thing the novelist can with any certainty call exclusively his own is in the inside of his own skull." 

With the development of generative AI, we are obviously facing another and quite similar identity crisis in the history of the arts. Now—not only can filmmakers do a better job of simply "telling a story" than the novelist; but AI language models are more efficient even at generating prose. Regardless of what we want to believe, AI may soon even be better than human brains at things like "the precise use of language" and the "exploitation of the technological fact of the book" and the "explication of thought." 

That knocks out several of Johnson's examples of what the novel could do that a film can't. AI can do just about anything with the combination of words and text that a human being can do—and if we tell ourselves that it can't, we probably are not paying attention to the latest models or being realistic with ourselves about how rapidly they are progressing. 

What then is left for the novelist to do in our era? What can the human writer now "call exclusively his own," as Johnson put it? 

Well subjectivity, still, certainly. AI may be able to write a steam of consciousness novel as well as anyone. But it cannot write the stream of my consciousness. Because it has not lived my life. It may be as good as any human at inventing fictional human lives; but it doesn't know any real one at first-hand. 

This is why B.S. Johnson's turn toward writing only about one's own true experience—to make novels out of autobiographical fact instead of imaginative fiction—feels so prescient as well. This is a domain that remains each author's sole province—their own lives. And so, we must write of that. "I am not interested in telling lies in my own novels," Johnson writes. Indeed, "the two terms, truth and fiction, are opposites," he points out (with perhaps a bit of willful pig-headedness for the sake of making his point). 

As Robert Lowell once rhetorically asked: "Why not say what happened?" And indeed—why not? If many a writer or poet has been said to have come up with nothing but disguised autobiography, why should this be a reproach? Perhaps it's true that Johnson's approach confines us to what Lord Byron called the "gloomy vanity of drawing from self." But what else is there? Why else should we write—when experiences other than our own, fictional ones, can be dreamt up by a computer just as well? 

Johnson's other reasons for writing survive equally well in an age of AI. He writes, he says, "because I have something to say that I fail to say satisfactorily in conversation, in person." He adds: "there are things like conceit, stubbornness, a desire to retaliate on those who have hurt me [...]" and suddenly one is reminded of Orwell's own quite similar list of reasons to write—which starts with: "Sheer egoism. Desire to seem clever [...] to get your own back on grown-ups who snubbed you in childhood," etc. 

These motives to create prose still stand perfectly well—one sees—even in an age of AI. A large language model may be perfectly well able to generate convincing combinations of words. But it cannot settle personal scores for you. Nor can it make you feel particularly clever, since it is doing all the heavy lifting. "Sheer egoism," perhaps. "Gloomy vanity." But without egoism, whence art? 

Art forms have survived and endured before by taking a turn into ever greater subjectivity. Why could we not survive by doing the same again? Why could we not retreat one step further into that redoubt where the machine cannot follow us: the hollows of our own skulls? And if this seems self-centered—well, the reason we record our own experiences accurately is surely for the use and enjoyment and instruction of others, ultimately. 

There is one other thing, however, that occurs to me as the unique preserve of human creators—one other thing that we can "exclusively call our own," namely: bad writing. Clumsy prose. In an age of language models, AI can increasingly do the job better than us of writing prose with polish. And so, we humans may have to lean further into writing that is awkward and ungainly. 

I used to be embarrassed by these moments in my own writing when they appeared—now, I begin to take a perverse pride in them. They are marks and badges that a human has been here. 

I once would have felt strange ending a piece like this on a dangling afterthought such as this one—by introducing this additional point about bad writing without a strong structural connection to all that had gone before. Now, I leave it here with pleasure. Perhaps I should write more posts on this blog that are fundamentally unfinished, imperfect, and full of loose ends. That will be the form of writing that remains the most human, after the new technology had remade the art of today.