In recent years, however, this long-standing critical consensus has come under assault. Thanks to brewing culture-war controversies around race and cultural appropriation, as well as recent anxieties about ChatGPT replacing Hollywood screenwriters, questions concerning who writes and who gets to write have returned to the center of mainstream discussions of literature and film.
It may seem as if these are separate concerns, but ChatGPT and identity politics are two sides of the same coin — both represent bankrupt versions of what literature is and is meant to do. We’re restricting what authors can write on the basis of their identity, and at the same time threatening to dissolve literature into machine babble. These two cultural forces are threatening the crucial function of fiction in our society.
Although we have grown used to the idea that actors should share the racial identity of the characters they portray on screen or stage, things get tangled when this notion drifts into the realm of literature. The backlash against Jeanine Cummins’s 2020 novel “American Dirt,” which generated explosive controversy owing to its white author’s depiction of Mexican immigrants, is only the highest-profile incident in these debates. Richard North Patterson recently courted much the same racial controversy with his new book, “Trial,” which features a black Georgia teenager as its main character.
Half a century after Barthes declared the death of the author, fraught conversations about literature and racial identity have complicated any efforts to consider writers in isolation from the characters they write.
Though it flies under the banner of progressive political sensibilities, this new racial orthodoxy — that white authors must write white characters, Black authors Black characters, and so on — assaults the very idea that it is possible to imagine human lives unlike one’s own.
One can do this poorly, of course. Some cross-identity depictions will inevitably be clumsy or even racist. But they should be judged on their literary merit. Instead, what we have now is “antiracism” by way of literary segregation — what Patterson recently called “redlining literature” — that reduces authors to mirrors who can do nothing but reflect their own lived experience. In this equation, readers are reduced to racial tourists.
This straitjacket betrays the promise — and premise — of fiction.
At the same time that these arguments over race and writing are producing growing discord about what it means to author literary fiction, Hollywood screenwriters are confronting another crisis of authorship: Human writers run the risk of being erased altogether.
That is the worry at the heart of the ongoing WGA and SAG-AFTRA strikes, as anxieties about ChatGPT replacing screenwriters have riled Hollywood. Experimental AI-produced trailers and commercials are already circulating — and yielding deeply unsettling results, as fireballs explode over backyard grill parties and faces seem to melt on screen. If tools like GEN-2 (which creates storyboards and even video from prompts) come into widespread use, this kind of weird effect could take the place of human creativity.
Most public conversation about the writers’ strikes has framed the threat of AI as a labor issue, with a few critics also lamenting that ChatGPT will likely produce television just as bad as the many bad novels it has produced.
What this discussion misses, however, is that what is at stake in this controversy over AI authorship is the same thing that is at stake in debates over authors and race: the very meaning of writing and the reason we consume it.
At the core of literature, television, and film is a kind of contract with the reader or viewer: Behind this novel or TV show, there is a human being like you from whose mind the work sprang. This contract holds that regardless of race or creed, regardless of differences in language or national origin, writers and those who consume writing can meet on a common ground and learn something about what it means to be human.
Writing produced by ChatGPT violates this basic contract, but its effect is the polar opposite of the identitarian straitjacket. Where rigid conceptions of race narrow our imaginative capacities, AI expands them — too much.
AI systems like ChatGPT rely on massive datasets. A trillion words, in which many novels were included, were fed into one version of ChatGPT for training. The late Cormac McCarthy famously observed that “books are made out of books,” by which he meant that anything an author writes reflects the influence of the previous authors they have read. The novel is never the product of only one human even if it has only one author: It reflects back to us every other genre, everyday speech, a panorama of our society. In this sense, ChatGPT might seem the ideal novelist, because the AI system has read almost everything. Yet while we would never accuse a novelist of being too well-read, this is precisely the problem with ChatGPT.
Attempts to use AI to write fiction are still nascent but proliferating. Stephen Marche’s experimental ChatGPT novel “Death of an Author” is a compulsively readable Raymond Chandler-style whodunit that plays ironically on Barthes’s notion. But even though Marche wrote “only 5 percent,” he says, of the words in the text, his tight control over the plot, sentences, and structure makes it mostly just . . . a novel, one that still reflects the guiding hand of a human being. Barthes’s assertion that the author is dead and Marche’s title aside, ChatGPT-composed novels resurrect the author too.
Sort of.
If racial identitarianism in literature shrinks human imagination, AI in fiction and film — because it has read more than any human could — expands imagination but does so in a way that breaks the fiction contract.
That’s because ChatGPT is the “web of citations” that Barthes wrote about in his essay. It is capable of knitting together new stories out of patterns of text it finds on the web, literally making McCarthy’s “books out of books.”
But fiction is an organized reflection of human language and culture. Because ChatGPT it is influenced by everything, it is also, then, influenced by nothing in particular.
Endless training, no taste.
The result is a distorted mirror of our collective unconscious, boring, ugly, impossible to ignore — but it is not fiction. It is instead a form of nonfiction disguised as a novel or a film, the whole web of all the citations — but it is not what we call story.
In the 21st century, the author isn’t dead. The author is a zombie, caught up in culture wars and quasi-dystopian digital progress. Debates over who gets to write what are not just debates about racial representation and cultural appropriation. They are not just debates about labor, AI, and automation. They are debates — occurring at a moment when the humanities are in retreat — over the most basic stakes of writing as a human endeavor.
If the arts and humanities are to have any hope of surviving the lean years to come — years that we predict will be defined by increasing political and racial balkanization, institutional defunding, and the AI-facilitated obsolescence of millennia-old modes of expression like the essay — we will need fiction in its classical sense: unrestricted by the identity of the author but also resolutely human.
Tyler Austin Harper is assistant professor of environmental studies at Bates College. His writing has appeared in The New York Times, The Washington Post, The Atlantic, and other outlets. Leif Weatherby is associate professor of German and director of the Digital Theory Lab at New York University. His writing has appeared in such publications as The New York Times and the Daily Beast.
