I stand before this honourable court as a counsel for African Voices. I pray that this honourable court consider my positions and grant my prayers.
I remember that last year, A white man took to twitter (now called X) to tell us that the use of the word delve by an African indicated that Artificial intelligence has been employed in writing.
My first reaction to this was shock. I grew up in a place where the word “delve” was thrown around by my lecturer almost all the time.
“Let's not delve into that for now”
This is a common phrase and you'll agree with me on that.
So imagine my surprise when someone took that ordinary word and slammed AI on it. I wonder what he'll do when he hears “nincompoop”, “brawl” “infinitesimal” … and the likes. I will even consider my vocabulary limited in the wide sea of words I hear and come across everyday.
There is this peculiar crime for which an African writer can be prosecuted without committing an offence. It is called sounding too intelligent.
A sentence arrives with impeccable grammar, and somebody reaches for a detector. A young writer uses juxtaposition, and suddenly her vocabulary requires an alibi. She builds an argument, employs a metaphor, remembers a proverb her grandmother might have used, and the jury begins to wonder whether she has been assisted by a machine. The African mind is permitted to be expressive, but there are limits to how convincingly it may demonstrate its education, if not it starts being questioned, scrutinized and placed under the lens of what is expected.
The accusation is almost amusing until one considers what it means. We have arrived at a point where a writer may have to defend not merely what she has written, but her capacity to have written it. The burden of proof has shifted. Instead of asking whether a piece is original, we ask whether its author looks, sounds or writes like somebody who could have produced it.
I find this troubling for a reason that extends beyond artificial intelligence. Before machines learned to write, the world had already developed an interesting habit of deciding what African intelligence should look like. AI has not invented that prejudice. It has simply allowed it to be more expressive.
In her celebrated TED talk, The Danger of a Single Story, Chimamanda Ngozi Adichie recounts how a professor considered her novel insufficiently African because her characters were educated, middle-class people who drove cars. An American roommate, meanwhile, was surprised that she spoke English so well. Both encounters were different expressions of the same assumption: Africa had already been imagined, and anything that contradicted that imagination required an explanation. Adichie's characters had committed the offence of being ordinary people rather than participants in somebody else's idea of African suffering.
What interests me is how familiar the accusation has become. The professor questioned whether her characters were African enough. The roommate questioned whether her English was Africanly possible. Today, AI detectors now mirror the suspicions that have long been held.
We can see that the most dangerous thing about artificial intelligence is not that it can imitate human expression. It is that we might begin to use its existence as a reason to distrust expressions we have never properly learned to recognise.
Consider the evidence. In 2023, researchers affiliated with Stanford University published a study examining seven widely used AI detection tools. They found that the detectors incorrectly classified more than half of the human-written essays by non-native English speakers in their sample as AI-generated. The average false-positive rate was approximately 61.3 per cent. The study did not establish that African writers are uniquely targeted, nor did it test African literary writing. Its findings were based on TOEFL essays by non-native English writers and essays by American eighth-grade students. Nevertheless, it exposes a serious problem: linguistic patterns associated with careful, less varied or predictable English can be mistaken for machine authorship. Even more revealing, the researchers found that modifying the vocabulary of the human-written essays significantly reduced the rate of false accusations.
There is something particularly unsettling about that finding. A person can write in a manner that has been taught, corrected and rewarded by educational institutions, only to have the resulting discipline interpreted as evidence of artificial production. The research is not proof that every accusation against an African writer is prejudiced. It is proof that the instruments of accusation are not beyond error.
I wonder what happens when a young African writer learns that the safest way to sound human is to sound less educated. Should she deliberately break a sentence, abandon a carefully selected word or make her argument less elegant to reassure an audience that has already decided what her natural voice ought to be? Must the writer introduce grammatical imperfections as proof of life?
We should be careful here. Not every accusation is prejudice, and not every beautifully constructed sentence is evidence of originality. Artificial intelligence can produce convincing prose. Writers can misrepresent machine-generated work as their own, and originality matters. A literary community that refuses to confront this reality would be protecting neither its writers nor its literature. But the existence of dishonest writers cannot justify treating every accomplished writer as a potential fraud.
The question is not whether AI can write. It can. The more important question is whether we have become so preoccupied with what a machine can produce that we have forgotten the extraordinary things human beings have always been capable of producing.
This is where I return to the phrase native intelligence.
Long before an algorithm could predict the next word, somebody had to teach a child the meaning of words. Long before a machine could generate a proverb, somebody had to observe life closely enough to make one. Our elders did not require a language model to understand that a child who washes her hands before eating is practising wisdom, A child who eats with both hands is nothing but a fool and that a tree with deep roots does not fear every passing wind. They had names for things, explanations for events, and philosophies compressed into sentences that could survive generations.
A Yoruba proverb does not become less intelligent because it is short enough to be remembered. A saying in Igbo does not lose its philosophical weight because it is delivered over a bowl of food. A Hausa folktale does not become inferior because its wisdom travels through storytelling rather than an academic journal. These are not failed attempts at Western intellectual traditions. They are expressions of knowledge with their own histories, methods and purposes.
Native intelligence, to me, is not an intelligence that rejects everything foreign. It is the ability to know something without first requiring somebody elsewhere to certify it. It is the capacity to observe one's environment, interpret experience, preserve knowledge, challenge assumptions and create meaning. It is intelligence that knows where it comes from without being imprisoned by that knowledge.
It is intelligence passed down from our fathers father to them and now to us.
And if we are to discuss African voices, we must first abandon the temptation to speak of an African voice as though the continent has one mouth.
There are thousands of languages, countless dialects, different educational histories and literary traditions across Africa. There is no single rhythm that can contain us. The African writer who composes in formal English is no less African than the writer who writes in Pidgin, Yoruba, Igbo, Hausa, Swahili or any of the continent's other languages. The woman who writes quiet literary fiction is no less African than the man who writes political satire. We are not obliged to sound alike to belong to the same continent.
In fact, some of our most significant literary achievements have come from writers who refused to sound alike.
Amos Tutuola wrote The Palm-Wine Drinkard in a distinctive form of English that departed from conventional grammatical expectations. Ken Saro-Wiwa's Sozaboy employed what he called rotten English, a deliberately constructed mixture shaped by Nigerian Pidgin and the speech of its protagonist. Chinua Achebe argued for an English capable of carrying African experience, while Ngũgĩ wa Thiong'o made a powerful case for writing in African languages as an act of cultural and intellectual liberation. And in more contemporary literature, we see what Abi Dare did with “The girl with a louding voice” Their approaches were not identical and their disagreements did not make any of them less African or more African than the other.
That is the point. African literature has never been a uniform performance.
Pidgin, in particular, deserves more than the occasional appearance as comic relief. It is a language of negotiation, intimacy, satire, resistance and everyday intelligence. It can communicate an idea with a directness that formal English sometimes spends an entire paragraph trying to achieve. It carries the humour of the street, the creativity of people who have made a language their own, and the ability to express complex experiences without seeking permission from grammatical conventions.
A person who says, "No be every open door be opportunity," has not necessarily made a less sophisticated observation than someone who writes an entire paragraph about discernment. The two may simply have chosen different roads to the same destination.
Our proverbs, too, are more than decorative additions to otherwise serious writing. They are intellectual inheritances. They carry moral reasoning, historical memory, social criticism and philosophical observation. Sometimes, an entire argument has been sitting in the mouth of an elderly woman long before somebody discovers a suitable academic terminology for it.
We must also be careful not to turn these traditions into another restrictive definition of authenticity. African writers should not have to insert a proverb into every essay, translate every cultural reference or make every character speak in Pidgin to establish their identity. That would be another form of the single story, only this time dressed in cultural pride. An African writer should be free to write about heartbreak, astronomy, corporate mergers, imaginary kingdoms, loneliness or the price of tomatoes without submitting a cultural explanation for every choice.
Our plurality is not a problem to be resolved. It is the evidence.
But what, then, makes a voice truly yours?
I do not think ownership of a voice can be established by vocabulary alone. It is not determined by whether a sentence is unusually beautiful, whether it contains an African proverb, or whether it sounds sufficiently different from something a machine might produce. Machines can imitate our metaphors, reproduce our idioms and arrange our familiar expressions into convincing paragraphs. They can learn the shape of our sentences, but resemblance is not the same as authorship.
A voice becomes yours through the relationship between what you write and what you can account for. It is in the experiences you bring to a subject, the choices you make, the questions you ask, the details you notice and the decisions you defend. It is in the intellectual labour behind the finished paragraph. Two people may write about the same subject, use the same language and arrive at similar conclusions, yet produce entirely different works because they have approached the subject through different lives and minds.
This does not mean that every writer must have personally experienced everything they write about. Imagination is not a crime against authenticity. A novelist need not have committed murder to write a convincing murderer. A writer need not have travelled to every country to imagine its landscapes. The point is that writing involves a series of human decisions, and originality is better examined through those decisions than guessed from the appearance of the finished product.
This is also where writing communities must become more thoughtful. We cannot protect originality by making suspicion our primary editorial method. A detector's score should not become a conviction, particularly when the reliability of the instrument is in question. Communities can ask writers to retain drafts, notes and revision histories. They can invite writers to discuss their creative decisions, explain their research and disclose the use of AI where it materially contributed to the work. They can establish transparent rules about what assistance is permitted and distinguish between using a tool for proofreading and submitting machine-generated writing as one's own.
Most importantly, a writer should have an opportunity to respond to an allegation before being judged. Evidence should precede accusation, not the other way around.
There is a distinction between protecting the integrity of literature and protecting our assumptions about how literature ought to sound. The first is necessary. The second is dangerous.
I also think we must resist the temptation to make AI the sole villain in this conversation. Artificial intelligence is a tool, and tools reflect the choices, data and assumptions involved in their development and deployment. It can help a writer explore an idea, identify weaknesses in an argument or overcome the intimidation of a blank page. It can also be used to manufacture work and misrepresent authorship. The ethical question is not simply whether a writer has encountered AI, but what role it played in the creation of the work and whether that role has been honestly represented.
We should not demand that writers pretend technology does not exist. Nor should we allow technology to make the human imagination appear suspicious by default.
I am particularly interested in this because I am a young Nigerian writer. There is a peculiar vulnerability in knowing that your words can be examined for evidence of an intelligence you are expected to possess but not display too convincingly. You want to write well. You spend time searching for the right word, rearranging sentences, deleting paragraphs, returning to an idea because the first attempt did not quite capture what you meant. You learn, read, observe and revise. Then the finished work arrives, and somebody wonders whether you had help. I have personally sat with a work and written it with nothing but my brain and two hands. I was quite surprised when a peer looked at it and told me that I had definitely used AI.
The accusation does not merely question the source of the words. It threatens to make the labour behind them invisible.
I would hate for a generation of African writers to begin editing their brilliance out of their work because they fear being mistaken for machines. I would hate for us to become so concerned with proving that we are human that we forget to be imaginative. There is already enough pressure on African writers to explain their identities, justify their subjects and satisfy the expectations of readers who have never met them. We should not add the burden of deliberately writing below our abilities.
A rather disgruntling irony is that the machine itself has learnt from us. Language models are trained on vast quantities of human-generated material, including writing from people whose names, labour and cultural contexts may never be acknowledged in the resulting systems. The machine's ability to produce language is not evidence that human language has become artificial. It is evidence of how much human language there was for it to learn from. The irony is not that a machine can sound like us. It is that we might begin to doubt ourselves because it can.
I return, then, to the courtroom.
If the charge is that an African writer has written too well, I plead not guilty. If the evidence is a vocabulary that exceeds expectation, I ask that expectation itself be examined. If the prosecution insists that the work sounds too polished to be human, I would like to know who established the permissible limits of human excellence, and why African writers were not invited to the meeting.
We must hold writers accountable. We must protect originality. We must learn to use emerging technologies without surrendering the values that make creative work meaningful. But we must also understand that a writer's voice cannot be reduced to a statistical pattern, and a continent's intelligence cannot be measured against a machine's ability to predict its next sentence.
The African voice is not a specimen to be authenticated. It is not a performance designed to satisfy an audience's memory of Africa. It is a living, changing collection of voices, sometimes familiar, sometimes foreign to one another, sometimes speaking in the language of our ancestors and sometimes inventing new ways to speak.
And if a machine can learn to sound like us, perhaps the more urgent task is to ensure that we never learn to sound less like ourselves.
In the words of Chinua Achebe :
Let no one be fooled by the fact that we may write in English, for we intend to do unheard of things with it.
The counsel for the accused has finished speaking. The court may now examine its assumptions and consider my prayers.
Thank you.
6





