I.
Here’s a sweeping and oversimplified and incomplete generalization: we can encapsulate the entire history of literature as a gradual narrowing of perspective. Whether you’re examining a Western tradition, the four great masterworks of Chinese literature, centuries of texts from the Middle East, or the literary classics from pretty much anywhere, we, as humans, have been slowly opening the aperture on the lens, with our literature now providing us with a view that’s closer but sharper. Our earliest stories might be described as distant, with a narrator that stands above the action and looks down, the tale collective, social, suitable for a lyrical epic; while our more contemporary stories thrust us into the individual and the particular, the narrator somehow closer, the story intimate, personal, even familiar.
One way to consider Athenian drama and Homer, for instance, is through a narrative voice that’s just off stage, with the storyteller breathing life into the characters while we watch, the perspective omniscient, all-encompassing. But that’s less common as the centuries pass, and there’s no shock once we start detecting hints of a more contemporary individual in Dante and Shakespeare or in the writing of Montaigne. The preponderance of fictional stories still resemble more we than I, and most storytellers still remain in the distance—yet they are starting to step closer to the stage.
Eventually the shackles that restrain our most classic narrators drop, and we end up with an explosion of stories that are shaped in a manner that modern readers would recognize—with Flaubert and Austen and Dostoevsky just a few notable names that hurdle past the sociological and into the psychological, as our old stories about groups and archetypes become—or expand into—stories about individuals. Of course the vanguard pushes the lines further just as they appear, and somehow, once again, nudges the perspective even closer, so that in the writing of James Joyce and Virginia Woolf we’ve gone so far inside that we’ve jettisoned any notion of narrative, leaving us with just the individual, with the writer’s voice, with the inhabiting of a sensibility.
And in the last hundred years of literary pivots—including many attempts to quash this focus on the individual—there’s a reasonable argument that, in general, the tendency has been a continued narrowing. The aperture of the lens has opened wider: the details continue to sharpen while the scene continues to narrow. This trajectory has taken our literature from the objective and into the subjective—how a story is told and who tells a story is the actual story in contemporary literature. Which means that nearly all conversations about literature today focus on narration and voice and perspective. Nobody should be surprised by the popularity of auto-fiction this century, nor is there anything shocking about the ubiquity of memoirs that include fictional characters, or even in all the controversies about who has the authority to write particular fictional stories—as we’ve shrunk every sentence to a mark of identity and a declaration of perspective.
II.
Consider three celebrated writers and three notable essays from last century: James Baldwin’s Notes of a Native Son, a poignant, unforgettable meditation about Baldwin’s father, hatred, the events of 1943; Joan Didion’s The White Album, a subtle, serpentine recollection of events in 1960s California; and David Foster Wallace’s A Supposedly Fun Thing I’ll Never Do Again, which describes, for some reason, Wallace’s experience on a cruise ship.
Unless you’re feeling particularly capacious, these three essays have little in common at first glance, with the first rather crudely and unfairly summarized as a personal story about race, the second about detachment and violence, and the third about entertainment. As to why these three essays have persisted for so long—each becoming the title essay in larger collections, each adored by countless readers—isn’t obvious if you examine those individual subjects. There are, after all, plenty of books about race, violence, and entertainment, and there are certainly more comprehensive and factual sources on other shelves. Baldwin doesn’t deliver any supporting facts, some readers might notice; Didion provides a few dates, crucial events, but we’re mostly lost in her confused mind amid the swirl of 1960s California; and Wallace doesn’t even attempt to drift from his own haphazard and frenetic thoughts, with his non-fiction account of the cruise coming from, strangely, an unreliable narrator. So why turn to these essays? Why read these writers on these subjects?
The obvious answer is that the subject is secondary, despite how some people may discuss these writers. If you’re merely interested in the subject—in facts, an argument, the history—then you’re looking for a list, or a textbook, you’re looking for an encyclopedia rather than a narrative. Which isn’t the reason that we remember our greatest writers. Nor even the reason that we find subjects that we don’t even expect—like the story of a cruise ship—worthwhile to read. What makes these essays striking, instead, is how they each inhabit a sensibility, how each essay comes with a incisive, obsessive eye for detail, along with an uncanny capacity for articulating what’s internal—Baldwin’s essay isn’t remembered because it is about race, it is remembered because it asserts Baldwin’s struggle and notions about race; Didion didn’t write the most comprehensive or even factual account of the 1960s, but her essay offers the only method of perceiving those years through her eyes; and Wallace takes, well, a rather banal subject and makes it interesting only by infusing every single detail with his voice.
III.
Here is OpenAI’s ChatGPT answering a prompt about the architecture in Bucharest:
The architecture in Bucharest is a captivating blend of styles, featuring a mix of elegant Belle Époque buildings, communist-era utilitarian structures, and modernist designs. The cityscape is defined by its diverse architectural heritage, showcasing influences from various historical periods and a growing emphasis on contemporary, sustainable construction.
Now that might be a reasonable and useful description, it is at least factually correct, though it contains a fundamental flaw, one that every English teacher, I hope, would spot immediately, and one in which every English student should also spot: who is speaking? Where are these sentences written from? Other than some arbitrary grammar rules that make these complete sentences, the passage is indistinguishable from a list of facts. This is trivia in sentence form: disparate facts that are combined into a narrative, but that exist without a foundation, without a perspective, which is why the reader feels a hollow, passionless sensation—it is bland enough to be tasteless. There’s no position for the reader to stand because the sentences don’t originate from anywhere—you can’t subsume yourself within the perspective of these sentences because they, remarkably, don’t contain a perspective.
Bucharest's architectural tapestry unfurls like a layered narrative, where the ornate grace of neoclassical facades dances with the stern, concrete echoes of a bygone regime, all under the watchful eye of ever-evolving modernist contours. Each edifice, a page in the city's tale, weaves a complex, poetic mosaic, revealing the eternal interplay of tradition and innovation on this urban stage.
Here is the same prompt but, this time, GPT was asked to write in the style of Vladimir Nabokov who, incidentally, never visited the city. Now I will grant that the diction and flow of each individual sentence is probably a better facsimile of Nabokov than your average Nabokov impersonator, so that’s at least noteworthy. But the space between the two sentences is where the flaws emerge. The limits of a disembodied machine trying to evoke architecture appear once you have two sentences, as it reveals a lacuna that would probably give Nabokov a fit: there’s no trajectory in these sentences. Pick up Lolita, or Pnin, or a grocery list scribbled by Nabokov, and you’ll spot—in any two sentences—an upward or downward slope, a moving of perspective, a trajectory that takes you from a beginning to an ending. Yet that’s missing from these sentences. You can read those fifty-nine words and almost miss that you haven’t traveled anywhere, and that you aren’t viewing the city from an actual eye. You could, in fact, flip sentence number one with sentence number two and nothing would really change—there’s no point of view, it is just a ceaseless, neutral, present tense, Nabokov replica.
In Europe's Shadow: Two Cold Wars and a Thirty-Year Journey Through Romania and Beyond is a fascinated, and personal, book about Romania by Robert Kaplan, who visited the country for the first time in 1973, again in the 80s, then continued to take more trips after the collapse of the government in 1989. Although I can’t quite excerpt a single passage to exhibit how Kaplan describes the architecture in Bucharest—as the entire book is itself a long reflection—I can offer a taste of his narrative, which is always evocative, lively, and personal:
I TOOK A WALK in Herăstrau Park, not far from Geoană’s office, where amid the little stone bridges and flower beds, while passing briefly through Bucharest…
With a clear delineation of changes, of how, to his eye, the city transformed over the decades:
Bucharest in 1981 had a theme, a pattern: that of Communism—somber, formidable, and suffocating in the silence it inflicted upon the streets and boulevards. Bucharest in 2013 had no theme, no great urban plan of the kind that had prepared monumental cityscapes such as late-nineteenth-century Paris and Ringstrasse Vienna.
Although Kaplan does give the reader a detailed history of Romania, and also describes the architecture in Bucharest—explaining the features and style of the most notable buildings in the city—the sentences never leave his perspective. It isn’t a mere factual account, as we read about, instead, Bucharest as interpreted by a single person, the city vibrant, shifting, alive, with the narrator’s voice guiding our path, even when that narrator is at home, listening to music, and pondering his memories:
Enescu’s music makes me think of, among so much else that I distinctively associate with Romania, the pure vegetal dyes and biblical settings of the painted monasteries of the Bukovina, and the original mix of Oriental and Occidental styles that constitutes the unique enchantment of Brâncoveanu architecture. I think of the earthy and vigorous smell of ţuică with its evocations of interbellum Bucharest café music.
There’s a difference between the clear, correct, and bland directions that your phone provides about a new city—and directions that include a memory, a description, and some advice, because you’re listening to the voice of someone who just visited that city.
IV.
A great essay about architecture isn’t an essay that describes a few buildings and a little history. It is one that describes buildings and history from a distinct, compelling voice, one that the reader finds alluring, particular, notable. Which doesn’t mean that generative text models are condemned to forever lag behind—only a fool, having read no history, would make predictions about impossibilities. Whether the processing occurs on silicon or in neurons probably doesn’t really matter, but it probably does matter whether there’s the potential for a disembodied perspective—whether there’s a distinct, autonomous voice from which the sentences are told.
Even when you’re curious about history, the root of your interest is deeper than any particular event: to read about World War II is, yes, to learn about World War II, but only on rare occasions does your curiosity begin with a desire to know facts. The trigger for your curiosity is always deeper, more human, and the event—however consequential—is valuable in what it reveals about those more profound desires. If that weren’t true, then a long list of facts about World War II would be fascinating, there wouldn’t be any need for narration, for an account that comes from a unique perspective—a unique perspective would, in fact, distract from our supposed interest in events. Information is insufficient when you actually want what’s most human.
And the importance of inhabiting a perspective doesn’t change if you alter the medium, or if you shift from fact to fiction. There’s plenty of arguments about the importance of stories, and why we’re obsessed with hearing and watching and even inhabiting the stories that surround us, but the surface elements—the actual stories—are often the least important aspect of our interest. Shakespeare stole most of his best plots, and nobody really cares because you don’t study Shakespeare to learn about events, just as you don’t read Dostoevsky because you simply want a murder story, nor do you read Austen because you want a factual transcript of a social drama. Any writing that overlooks this fundamental aspect of storytelling will forever miss the point.
Our desire is to get closer, to inhabit a more subjective perspective rather than one that’s objective. Writing that fails this basic test, that doesn’t permit empathy because there’s no perspective to inhabit, will always feel vacuous, disembodied, there will always be something missing, a lacuna that the reader senses but can’t quite name. Our need for connection is too strong—we don’t tell stories to learn mere facts, as we want our writers to, for just a moment, reveal what’s inside. We want our sentences to settle into a sensibility. To nudge us closer to the individual. It is the war as told by a person. Architecture as seen by a traveler. Until, eventually, all our stories, in some ways, become autobiographical, with every single genre—science fiction, romance, adventure, detective, literary fiction—now a subset of the panoptic genre of memoir.
Important. Tons of writers test the app, show the output, point at it and say, "See? That's bad writing!" But it's actually important to dig in and explain why, over and over and over again, because quite frankly few people recognize good from bad writing and fewer still know how to express why when they do.
Great post explaining everything that is fundamentally wrong with AI Writing. I have to say also as a photographer, that there's a lot fundamentally wrong with AI Images. One instance, is AI generated bird images in which the birds pictured are not the right color blue or red, etc...