Meet Your New Robot Co-Writer

From Esquire:

Some art forms welcome, even require, collaboration. After all, it is the exceptionally rare film or television show that gets made by a single person. Music, too, often literally demands the assistance of others. Even in these cases, though, there is a tendency to flatten the many into if not the one, then at least the few. Films—enormous undertakings costing millions of dollars, employing hundreds of people in numerous fields—have an entire theoretical construct organized around this very flattening: auteur theory. Emerging from the French New Wave of the 1950s and epitomized by the New Hollywood iconoclasm of the 1970s, auteur theory argues that the director is the sole author of a film, or the figure to whom we should attribute the work. The notion of a single figurehead was codified in 1978, when the Directors Guild of America added a provision to its bylaws, the “One Director to a Film” rule (Article 7-208) that can only be bypassed if, as an IndieWire column put it in 2022, “director duos… apply to the union’s Western Directors Council and make the case that they are lifelong collaborators; one-offs shouldn’t even try.” Lifelong collaborators—that’s a high bar to clear.

As for the many musicians and producers and technicians that are involved in, say, writing and recording an album, the front cover still tends to credit, by virtue of its prominence, a single entity. Fans sometimes even mine famous songwriting duos like Lennon and McCartney to figure out who really wrote “Happiness is a Warm Gun” (Lennon) or “Rocky Raccoon” (McCartney). We seem prone to narrow credit down to the fewest number of creators possible, despite what the bylines say.

But what of literature? Creative literary art forms, particularly fiction and poetry, are inherently solitary pursuits. There is nary a legacy of prosperous partnerships, and almost no tradition at all of enterprises with more than two authors. There are—of course—exceptions (some of which we’ll get to), but think of it this way: How many classic novels are written by more than one writer? How many collaborative novels have you read? How many are taught in schools? How many can you even name?

More importantly: why is this the case? Why aren’t there more literary collaborations? To be sure, in fantasy, science fiction, romance, horror, and mystery, there is a much stronger history of partnerships, but less so in so-called literary fiction. Novels written by groups are rare in any genre. In literature, we venerate the one over the many, and while it’s tempting to explain this away by citing the solitary nature of the practice—or, perhaps, the singular voice written texts seem to represent—I would posit that writing is not, nor has ever truly been, a wholly solitary act, and that collaboration occurs way more often than we like to admit or can even consciously acknowledge. But this tradition of encouraging solo art over teamwork—which in the past has led to the industry of ghostwriting and to numerous incidents of plagiarism—now has a new, fiercer, and much more insidious unintended consequence: the inevitable rise of fiction co-written by AI, and the flattening of literary storytelling.

I ask these questions because of the publication of Fourteen Days, a novel written by 36 authors ranging from John Grisham and Erica Jong to Tommy Orange and Nafissa Thompson-Spires. The project was overseen by Margaret Atwood and Douglas Preston, who both also contributed. Supported by the Authors Guild Foundation, Fourteen Days tells the story of a new superintendent of an apartment building on Rivington Street in the Lower East Side of Manhattan. The place is called the Fernsby Arms, which the super describes as “a decaying crapshack tenement that should have been torn down a long time ago.” The super tells the reader to call her 1A, an Ishmael for our journey through Covid, for as soon as 1A begins her job at the Fernsby Arms, which features “rotten” pay and a dingy basement apartment, the pandemic hits. The previous super lets 1A in on a little secret: the super can access the roof of the building, which provides her with stunning views of the emptied city. She hopes to keep her rooftop oasis a secret (for legal reasons), but soon the tenants discover this pocket of paradise. In the boredom and isolation of quarantine, they begin to connect with each other by telling stories. Fourteen Days is a kind of a miniature Decameron for this unsteady decade.

The stories alternate between humorous and poignant, personal and historical, dark and light. There’s an added component of wondering which of the 36 contributors wrote which part, a game I recommend readers not spoil by consulting the authors’ bios in the back of the book, where the contributions are itemized. It’s fun to try to link a story to its teller. Sometimes it’s obvious: I figured that James Shapiro, author of Shakespeare in a Divided America and The Year of Lear: Shakespeare in 1606, wrote the section about the Bard’s own encounter with a pandemic, but R.L. Stine’s brief addition wasn’t as easy to sniff out (though one fact should have clued me in). The Covid stuff—the uncanniness of empty streets, cheering for essential workers, toilet paper shortages, newly acquired hobbies—will affect each reader in their own way, but as for me, I didn’t particularly love being back in that atmosphere of novelty and monotony.

What struck me most, though, was the novel’s self-reflective theme of collaborative storytelling. The characters share themselves with each other through narrative.

. . . .

What Fourteen Days is not an argument for, however, is the beneficial results of literary collaborations (particularly group collaborations) as an artistic experience for readers—which is to say that as a whole, Fourteen Days is really bad.

. . . .

There is no doubt that the 36 writers whose names crowd the perimeter of the novel’s cover very likely found the experience of creating Fourteen Days to be fun, interesting, enlightening, profound, and instructive. Indeed, Fourteen Days is a portrait of, as well as an argument for, the vitality of collaboration. But there is also no evidence here that the greatness of individual writers is usefully mixable—somehow the math of literary creation functions inversely, where the higher the number of each factor in an addition or multiplication problem, the smaller the solution will be.

. . . .

The problem can be put this way: the act of collaborating is rewarding and enriching for the collaborators, but the results don’t usually hold up against the standard set by singular voices. As such, there is an enormous gap between the value of the act and the merit of the result. This divide has become newly relevant as our notions of what it means to collaborate, as well as our fixation on individual achievement, are now up against a force, a tool, and a readymade collaborator in artificial intelligence. And if we don’t amend our concept of human-to-human aesthetic cooperation, then we will potentially face an era of more collaboration, and yet somehow less uniqueness.

. . . .

If I am responsible for 99% of a finished piece, can I truly claim to be the sole creator of that piece? What about 82%, or 68%? When does it become dubious—or even unethical—to contain the numerous co-authors under the umbrella of one? Some might argue that these issues are essentially legal ones, as abstractions like “credit” are less about actual attribution than about profit-sharing and royalties, which result more from agreements made by the creators than honest accounts of their creations. But what if the situation were not a handful of people’s work being left out of the credits? What if it was something like hundreds, thousands, millions of contributors’ work not only going unacknowledged, but in fact remaining impossible to detect in a creation ultimately credited to no one?

Sean Michaels’s lovely and thoughtful novel Do You Remember Being Born? tells the story of a septuagenarian poet named Marian Ffarmer [sic], who accepts an offer from a major tech corporation to co-write “a long poem” with their artificial intelligence, named Charlotte. “Charlotte’s been trained,” Marian is told by an employee, “on a massive data set of poetry books and journals, on top of a basic corpus of ten million web pages. Two point five trillion parameters…”

Marian’s experience with the AI oscillates between awe at the poetry bot’s ingenuity and dismay at its algorithmic approximation of meaningful expression. At first, Marian can “taste the disjuncture between our lines, like chalk and cheese,” but then something changes, though Marian recognizes how difficult such shifts can be to accurately diagnose:

The software learned me to some degree, or I learned it, or else nothing changed at all except my posture toward its work: that instead of awaiting an obvious fit, hook and eye, I anticipated that band of friction, as a spade awaits the dirt.

The novel takes its time deftly and tenderly interrogating the nature of meaning, the inexhaustive dexterity of language, and our knack for finding thematic or linguistic connections where none were intended. Marian’s generosity and curiosity provide her with an openness about Charlotte’s literary abilities, while her intelligence and expertise refuse to permit mediocrity or meaninglessness. The delicate drama at the novel’s core comes from this complex inner conflict.

The conversations between Marian and Charlotte are written; Marian types into Charlotte’s software and hits the Proceed key. Marian’s typed side of the exchange is rendered in a unique font, while Charlotte’s responses are highlighted in gray, which usefully differentiates between the two and visually reinforces Charlotte’s fabricated isolation: Charlotte is literally kept in dull, slate boxes. But it performs another function: what does it mean when those gray highlights appear not in scenes featuring Charlotte, but moments in Marian’s personal life? In the novel, it’s a hint that Marian might eventually collaborate with Charlotte (or another AI) on more than just this one assignment.

Link to the rest at Esquire