There are thousands of articles written daily about artificial intelligence (AI). Some by humans, some by the machines themselves. Some by humans sounding more like machines, some by machines pretending to be human. It is the topic of the moment, and maybe of this generation. This is another one of those articles. But one that argues that in allowing AI to take the reins on communications and research writing, we give away one of humanity’s greatest conquests: the right to express ourselves on our own terms, rooted in our realities. 

This is especially relevant for the Global South. For centuries, the stories of the majority of the world’s population were hijacked by those who held power. Only through liberation, decolonisation, and education was protagonism disputed, as these movements lead with self-proclamation. It was a new era for the Global South to thrive through its own evidence and stories. What could go wrong?  

It is no coincidence that the rise of AI happens during a global crisis, with attacks on truth and evidence. Communications and research still serve as checks and balances to a world in chaos, sliding back to the same old monopolies of power. Of authority. Of narratives. Precisely because they protect knowledge as a public good, the threat of their standardisation through AI tools is so grave, and why the politics of writing matter now more than ever. After all, AI can write. But can it tell the stories of the Global South?

In defence of risk and language 

AI has changed our perception of writing. Transformed it into something procedural, objective, impersonal. Worst even: detached, predictable. Standard. More troubling still is that this flattens our self-expression, not just the writing. And what is the cost of standardisation in a world defined by its inequalities?

Automating writing also automates format, structure, vocabulary, priorities, biases. Tone. In generic formulas, language itself disappears, a sacrifice to the Goddess of Optimisation. Contrary to what we humans value, the machine rounds all the edges of style, culture and imperfections that stand in the way of frictionless, palatable content. 

Trained with the same old imperatives of the Global North, AI tools thrive on our complacency to wield power. To treat writing as if it is not a definitive act of choice, risk, and interpretation. As if mistakes and experimentation are not essential steps in human learning. There is value in the sheer act of struggling to select words, crossing out digressions, throwing away a first draft, a second, a third, a tenth. There is risk in selecting what to include, what to leave out, how to frame. Even bad linguistic habits have purpose – above all to be fully ours, stubborn reflections of ourselves and our cultures. AI will never capture that.

Language itself resists conforming to standardisation. Romance tongues often need 15% to 20% more words than English to express the same message. Across Asia, the Latin alphabet cannot match the depth of visual writing systems or scripts. In Africa, some cultures do not even rely exclusively on written format to capture culture, carrying themselves forward through oral tradition.

Meanwhile, AI is designed to write in simple, flat, predictable ways. There are no risks. Words are all in the right order, sentences do not vary length, paragraph structures are a cage. There is no deviation, no suspense – worse, no rebellion. There is no culture. But it is clear that language is not just a tool, but a whole worldview. And there is a fundamental difference between producing content, and producing meaning. There are risks we should be willing to take by insisting to write ourselves. When we do not actively choose the words that will define us, someone else (or something) will choose for us.

The downfall of expertise 

What results from the fact that anyone could write, even AI, is that anyone can also be an expert, even AI. This streamlining of thought in writing also means that neither communications nor research will require the same level of expertise. The bar is lower, so the high and low outliers are neutralised. Ultimately, AI changes who is recognised or validated as a knowledge producer. 

For the Global South, that has dire consequences. Limiting Global North frameworks were evolving – proposals in prescribed formulas, predetermined funding priorities, top-down agendas, and hierarchical partnerships. Now, we risk sliding right back. With AI providing a shortcut to the South’s expertise on demand, there are no incentives to engage in knowledge power-sharing. Nor incentives to financially value it. A convincing approximation, trained on uncompensated labour, will suffice. 

In this context, there are no efforts to share the power, only the knowledge. Neither to reach beyond tokenism, where representation is present but redistribution is not. This has serious economic consequences too. Without governance, AI will allow anyone to access the expertise, while barring the very experts from economic access, ascension or compensation. If these extractivist dynamics sound all too familiar, it is because they are. The pathway is carved: knowledge extracted from the South will fill up big tech companies, then return stripped of any context or meaning. AI platforms will harvest all the value of scaling up capability, with none of the accountability for compensation in return. 

Knowledge sovereignty, then, becomes crucial, and the Global South’s prime tool against re-colonisation in the age of AI. It enables the region to maintain ownership of its expertise during artificial intelligence’s new extraction economy.

Evading truth and accountability 

With AI diluting how we express ourselves and redefining who produces knowledge, it can ultimately change who defines truth and accountability. With writing delegated to a machine, authority and legitimacy then become outsourced rather than developed. 

Communications and research are first and foremost exercises in judgement. They ask from us an ability to trust inward as much as to look outward. When this judgement is weakened, humans lose touch with our own experiences, and AI decides the most optimised version of the truth. In the name of an illusion of further “objectivity,” we flatten the very subjectivities that make knowledge meaningful to us at all. 

Algorithmic consensus is not truth. Standardisation is not truth. Collective human knowledge is also not an absolute truth – but it is the product of our shared experiences and agreements about the world. It is indeed the closest we can get to the truth. It surely is not infallible, but valuable precisely for being collectively constructed and agreed upon as our social contract. 

The lack of one is the beginning of our downfall. With today’s AI-polluted and highly saturated information ecosystem, it becomes harder – not easier – to verify information and evidence. Mass deployment of doubt and confusion is in the best interest of big techs and of authoritarian governments, as it is sure to generate impunity. Hence, in communications and research alike, evidence needs interpretation, not just summarisation. With the surrender of this expertise, we can no longer even agree on truth, nor enforce accountability.

Caring as a political choice  

The challenge is posed, but what if we dare to care? 

When launched, AI’s Trojan Horse was presented as a tool for human improvement. It promised to take up tedious tasks to free up time for human experiences, including art, writing, family, hobbies. With time, AI has inundated those spaces too. By allowing this to happen, we show that we are people uncaring about everything that matters. That even the things most dear to us are not immune to our permissivity. 

But caring is a political choice. It is important to care about something, anything, and especially about what we put out in the world. We should care at least about craftsmanship, a defining human skill. We should care about writing, the very basis for our communications and research. We have to care – out loud, unequivocally – about making an effort for this care to be perceived. 

By not caring, we allow the colonialism of words by AI to reign unchallenged. The Global South is in a good position to lead this resistance, so we can overcome an even greater threat: the one of a captive imagination. The dynamics of AI, generating futures on our behalf, prevent us from even daring to imagine fairer, more expansive ways of existence. Yet the task of creation has always been ours to take, hence the need to choose, to risk and to write it ourselves. 

AI is here to stay. It is a new frontier, its fragile limits still written in sand. Knowledge sovereignty should lead the Global South’s AI governance agenda, preventing it from becoming a tool for self-colonisation. We can still shape it, by ensuring that we are the ones holding the pen of its rulebook. After all, the language AI generates is what defines them, just as the language we create is what defines us too. This is why caring about the politics of writing matters. In the age of AI, caring enough is revolutionary.

RELATED POSTS