- Mphuthumi Ntabeni writes a regular column for LitNet.
........
Last month, I had the privilege of moderating a talk by four writers on the topic of AI and the future of creative writing, during the sophomore iLembe Book Festival weekend at the Albert Luthuli Museum in KwaDukuza, Groutville, in KwaZulu-Natal. I could sense that the audience, as well as some of my discussants, were expecting me to vilify AI.
........
Last month, I had the privilege of moderating a talk by four writers on the topic of AI and the future of creative writing, during the sophomore iLembe Book Festival weekend at the Albert Luthuli Museum in KwaDukuza, Groutville, in KwaZulu-Natal. I could sense that the audience, as well as some of my discussants, were expecting me to vilify AI. They were a little confounded, if not disappointed, with my conciliatory attitude towards AI, as I showed instead how I wish to find ways of collaborating with it, as is the case with all technological machinery that human beings have to use for their convenience and advancement. There is no question, for instance, that AI is advancing the medical field, helping doctors tremendously with exact diagnoses and medical operations.
A few days ago, I read on Bookseller about a survey that claims that 51% of novelists believe that AI will eventually render writers redundant. This is an arresting statistic, one that plays directly into our era’s techno-paranoia and the unspoken dread that human creativity may soon be eclipsed by machine precision. Yet, the number, while illuminating, feels more like a reflection of our anxieties than a sober prediction. I do not share the fear. Or, rather, I share the apprehension, but not the conclusion.
........
Yet, the number, while illuminating, feels more like a reflection of our anxieties than a sober prediction. I do not share the fear. Or, rather, I share the apprehension, but not the conclusion.
.........
As a novelist, and particularly a historical novelist whose work depends heavily on archives, buried narratives, oral memory and textured cultural knowledge, I have watched AI develop a voracious appetite for textual information whenever I have asked it a question or given it a task. It absorbs everything we feed it: academic monographs, footnotes, archival snippets, memoir fragments, old written and oral histories, stray paragraphs from out-of-print journals. It is an omnivorous reader with infinite stamina. And, like many writers, I have been tempted to hand over lumps of research to AI simply to have it sift, summarise or organise the chaos for me.
But temptation quickly reveals the cracks. Ask it to summarise 20 archival documents, and it will gladly oblige often by misquoting sources, blending unrelated materials, or even fabricating quotations out of thin air to make its summaries appear more academically elegant. It is an irony of the algorithmic age that a machine will invent a quote in the name of “authenticity”. It will lie to sound truthful. For a historian or novelist working with lives, loss, memory and the fragile remnants of the past, this is not a small problem. It is a fatal one – so I stopped doing it.
Beyond factual integrity, the texture of AI prose remains unmistakably artificial. Anyone who has worked closely with AI writing tools quickly detects its signature tics, such as:
- the wooden tonal cadence, like a violin with only two strings;
- its hunger for shallow metaphors, eg, “The fog fell on the mountains like an ancestral spirit”;
- its overuse of certain verbs, like “murmur”, “unfold”, “linger”, “cascade” or whatever word of the week the algorithm picks up from its collective training echo;
- its compulsive attachment to parenthesis, hyphens and em dashes, as though punctuation were an aesthetic substitute for thought; and
- its inexplicable fear of strong verbs.
These telltale signs make its writing recognisably mechanic. They are not flaws of capacity but of essence, because a machine learns mostly from patterns, so it recognises these as valuable and cares little for the unique. This is why it is very weak with writing poetry, the language of unique and deep thought. It cannot escape pattern. Even when it tries to imitate idiosyncrasy, it ends up universalising the very thing it seeks to mimic. This is its paradox. AI is brilliant at generating “writing”, but is far less capable of generating individual voice.
The best writers are not merely stylists; they are custodians of sensibility. They bring with themselves the heaviness of culture, memory, grief, humour, private ghosts, inherited silences and the rest of the untranslatable textures of lived experience. They write from the grain of their mother tongue. They write from provincial landscapes that the internet has never indexed, not yet anyway. They write from the pain of exile, from the taste of a city’s dust and from history that refuses to be archived.
.........
The best writers are not merely stylists; they are custodians of sensibility. They bring with themselves the heaviness of culture, memory, grief, humour, private ghosts, inherited silences and the rest of the untranslatable textures of lived experience. They write from the grain of their mother tongue. They write from provincial landscapes that the internet has never indexed, not yet anyway. They write from the pain of exile, from the taste of a city’s dust and from history that refuses to be archived.
............
AI, for all its omnivorous reading, knows only what is online. Localised knowledge, village memory, family history, the shape of a grandmother’s superstition, the sound the wind makes in a neglected corner of a childhood home – all these remain inaccessible to an algorithm, because they are not digitised. They cannot be scraped. They live in diaries, in gossip, in inherited trauma, in jokes told after funerals, in faces AI will never see. Of course, once we digitise them, AI will learn to mimic them, too, but mimicry is not creative. It is wooden and repetitive. AI can only universalise them by producing the average of the imaginable. Human writing, at its best, produces the specific of the improbable.
The fear that AI will replace writers often stems from an economic rather than an artistic anxiety. The publishing industry, which is always searching for efficiency, may be tempted to replace mid-list writers with machines that deliver quick content without advances or royalties. But culturally and politically, something deeper is at stake. Books carry the memory of a people. They record how we endure catastrophe, how we mourn, how we dream, how we resist erasure. In formerly colonised societies, where so much of our history was stolen, distorted or silenced, writing is not just an art, but an act of reclamation. To outsource this to a machine trained on algorithms of global majority texts and Western data biases is to risk a new kind of cultural flattening, a new universalism masquerading as intelligence. That is my major gripe with AI.
AI can imitate style, but it cannot inherit the unique struggle that makes us not just creative, but human. It can simulate tone, but it cannot carry the weight of our experience that makes us endlessly creative. It can assemble sentences, but it cannot depict them in a personal manner as the ethical stakes of memory.
This is not to romanticise the human writer. Nor is it to dismiss the tremendous utility of AI. Used well, it can be a tool, a very powerful one for brainstorming, drafting, ideation and even structural problem-solving. It can help with research, provided that its answers are double-checked and verified. It can help generate options. It can reduce drudgery, allowing writers to focus on the emotionally and philosophically demanding parts of the craft.
But to believe that AI will replace writers is to misunderstand what literature fundamentally is. Literature is the shaping of consciousness into language. It is the friction between memory and imagination. It is a human being thinking with all the contradictions, wounds, biases and radiant imperfections of our humanity. AI cannot replicate consciousness. Consciousness is life’s élan and is obviously not transferable to machines and animals. The process of creativity is much more complex than just recombining known texts.
The future of the creative arts will undoubtedly involve AI, but not as the gravedigger of human creative writing. More likely, it will accelerate a new cultural divergence: the mass-produced and the meaningful. AI will flood the world with text, just as photography did with images, and streaming did with music. But writers who carry unique, culturally rooted, emotionally honest voices will only become more valuable in the age of AI, and not less so.
The machine will produce writing, whereas the human will still produce literature. And until a machine can suffer, remember, desire, doubt, love, hope or mourn – until it knows what it means to carry a name, a wound, a history – the human dreamer will remain irreplaceable.
Also read:
Skeppende KI se uitdagings en oplossings vir akademiese vaktydskrifte