Ten years ago, I found myself in Lusaka, Zambia, sitting at a table with a group of extraordinary writers. We had been invited to participate in a 12-day writing retreat by the Caine Prize for African Writing, considered by many to be the most important literary award for authors on the continent. Among our ranks at the retreat were four former winners of the prize (NoViolet Bulawayo, Okwiri Oduor, Namwali Serpell, and me), as well as past finalists and immensely talented Zambian writers. During those two weeks, we wrote and workshopped our stories, and in between—as we ate and gossiped, visited local schools, and read our work to the public—we discussed the great distance between the literature we hoped to produce and the kind of work the marketplace expected from us.
We were all beneficiaries of the Caine Prize, which was first awarded in 2000 to introduce African writing to a wider international audience. At the time, African fiction hardly existed in Western bookstores or newspaper book sections. In the years since, it had flourished, but many of the most successful and celebrated works circled the same themes: war and its aftermath, the scourge of poverty, the persecution and murder of queer people, the trials of immigration to the West, recapitulations of popular American or European stories in African settings.
In Lusaka in 2016, we wondered whether there was room for stories that did not hew to these norms, stories that followed their own idiosyncratic paths, stories that challenged and reconceived what “Africa” means. So many of us, after all, were living examples of the diversity and capaciousness of the term. Even as we pursued originality in our work, we recognized that certain stories (I described them as “accessible” in an essay for the Los Angeles Review of Books) would continue to earn a disproportionate amount of recognition and acceptance.
A decade later, the Caine Prize has become the setting for a different conversation about African writers and literary authenticity. Two weeks ago, the Zimbabwean writer Simbarashe Steyn Kundizeza was named this year’s winner for his story “The Blue Fighter.” Just five days later, the novelist and journalist Vauhini Vara reported in these pages that the AI detector Pangram had flagged almost the entire story as AI-generated.
The allegation quickly became fodder for arguments far beyond the Caine Prize: about the capabilities of artificial intelligence, the state of contemporary literature, and even the legitimacy of Black intellectual achievement. Vara’s story revived fears raised earlier this year, when three regional winners for the Commonwealth Short Story Prize were accused of using AI, that chatbots are already winning prestigious literary prizes by fooling multiple award juries. Critics also worried that contemporary literature had grown so stale that even a machine was capable of creating award-winning prose. But the racial dimensions magnified the controversy. Some people contended that AI-detection tools were being used to unfairly undermine Black and brown authors. For others, the accusations confirmed a long-standing belief that Black and brown writers—and, by extension, Black and brown people—are incapable of achieving prominence without substantial assistance, either from institutions that supposedly favor them over more qualified individuals or, now, from artificial intelligence.
[Read: Yet another literary prizewinner seems to be AI]
Lost in these debates is a quieter, more unsettling irony: After years of being encouraged by the literary ecosystem—editors, prize committees, magazines, critics—to produce work that is accessible to white audiences, African writers are now facing a new kind of suspicion about their authorship. In short, we have arrived at a moment when writing that conforms to accepted literary conventions—including, in particular, those associated with “African” writing—can also seem suspiciously contrived.
Kundizeza’s story provides a signal example. I found it moving and quite effective in moments, but I also instantly recognized—and predicted—many elements of its plot. The story is narrated by a boy in Zimbabwe whose father abandons his career as a civil engineer to pursue his dream of building aquariums and breeding exotic fish. While his father pours the family’s dwindling resources into the tanks and their expensive equipment, the boy’s mother struggles to keep the household afloat.
Perhaps you can imagine what happens next: The family moves to a poorer neighborhood, the mother finally confronts the father about his obsession, and, over his objections, she sells nearly all of his remaining fish tanks to pay the bills. Kundizeza’s story had many components of prize-winning African literature: a youthful narrator (my own Caine Prize–winning story, “Miracle,” featured one as well); the struggling father with big, impossible dreams; the responsible and enterprising mother who holds everything together; the backdrop of mounting economic hardship.
Kundizeza has denied using AI and has said that he drew on characters and settings from his own life in writing the story. I don’t doubt this, but one reason the story might strike some as AI-generated is that it so neatly encapsulates and deploys the tropes we have come to expect from prize-winning African fiction. If chatbots do anything especially well, it is exactly this: ingesting our preferred tropes and delivering them right back to us.
Readers who raise suspicions about AI usually flag patterns in prose. But there is another pattern worth considering. Chatbots have learned from massive quantities of human writing, including representations of Africa shaped by the literary marketplace. It seems reasonable to expect that if you ask a chatbot to write an “African” story, it will reproduce familiar conventions that have been rewarded for decades.
[Read: A turning point in AI writing]
That said, I don’t believe that a Pangram result should be enough, by itself, to condemn a writer. Earlier this year, the Commonwealth Foundation investigated allegations against several of its regional winners, examining drafts, notes, and time-stamped documents before concluding that AI had not been used to write their stories. In a recent New York Times essay, the Cambridge professor Doyle Calhoun noted that the five literary controversies involving AI that he considered most significant this year all concerned Black and brown writers. The Kundizeza accusation adds another. Whatever the outcome of the case, these controversies reveal a troubling predicament: Even as the literary world struggles to determine what constitutes AI-assisted writing, Black and brown writers have repeatedly been forced to defend the authenticity of their work.
One of the problems with adjudicating these situations is that no one has spelled out what, exactly, constitutes cheating. It is virtually impossible to open up a Microsoft Word or Google document without encountering AI tools that encourage you to rewrite your prose. If you accept that assistance and your work is subsequently flagged by a detection program as AI-generated, have you cheated? What if you drop a completed story into a chatbot text box and request edits? Established writers have the benefit of publishing their work in well-resourced periodicals where it is edited by multiple humans; the resulting article, regardless of how many sentences were drafted, deleted, or polished to a sheen by other hands, is credited to one person. Is it wrong for an emerging writer to seek similar assistance from artificial intelligence? Many people are quick to accuse individual writers of cheating, but technology companies have integrated AI assistance into so many parts of our lives, and the literary world has yet to agree on where assistance ends and authorship begins.
My greatest worry is that the arrival of AI has imposed yet another set of expectations on Black and African writers. Not only must our work satisfy the literary marketplace, but it must also distinguish itself from the machines that have learned to imitate its conventions. Over the years, the establishment has rewarded predictability and accessibility—the very qualities that now make writing suspect. As our Zambian retreat drew to a close a decade ago, many of us reached a conclusion that seemed almost impractical then but feels urgent and essential now: We had to ignore the noise and market signals and, to borrow from the South African anti-apartheid activist Steve Biko, write whatever we liked.