Why AI Fatigue Is Not About the Technology — And What Kenyan Creatives Can Do About It
A creative industries essay — Nairobi, 25th May 2026
By: Alex Gakuru
By: Alex Gakuru
In April 1812, a nineteen-year-old apprentice named John Booth lay dying in an English village, his body broken from a failed attack on a textile mill. The priests came for his confession. They wanted names. Booth whispered his famous last words — ‘Can you keep a secret?’ — and took the names of his fellow Luddites to the grave. But it is an earlier plea that reaches across two centuries and lands directly in the inbox of every Kenyan creative who has ever stared at a ChatGPT prompt and felt something between dread and disgust. ‘The new machinery,’ Booth said, ‘might be man’s chief blessing instead of his curse if society were differently constituted.’
Two hundred years later, we have our own machinery. It does not smash looms; it generates images, writes scripts, composes music, and mimics voices. And we are being told, from every corner of the internet and every pitch deck, that this is our chief blessing. That we should rejoice. That if we do not adopt, we will be replaced. That the ones who hesitate are Luddites. That they are normies.
The most important fact about artificial intelligence is that it does not exist. What exists are statistical models, each owned by someone, each trained on something taken from someone else, each priced to extract more than it returns. The sellers need you to believe the technology is the story, because if you look past it, you will see the oldest story there is.
There is an illuminating framework that cuts through the noise, broadcasting from Creatives Garage in Nairobi — home to Co-Lab X, AI x Tech, Artivism, and Market Access programmes. Every technology, from the plow to the large language model, is animated by three forces: money, power, and control. Strip the marketing, and what remains is a struggle over who pays, who plays, who rules, and who decides.
You can feel the fatigue in every creative space in Nairobi. The WhatsApp groups flooded with image generation experiments. The grant applications that now demand an ‘AI integration strategy.’ The workshops where someone stands at a podium and explains that generative AI is inevitable and you
must either get on board or get left behind. The subtext is clear: adopt or be marked as a normie. Adopt or be named Luddite.
At Creatives Garage, where Co-Lab X projects cross-pollinate disciplines, where the AI x Tech programme wrestles with exactly these questions, where Artivism wields creativity as political force, and where Market Access tries to build bridges between the artist and the economy — the frustration is real. It is not a frustration with the technology itself. It is the exhaustion of being told, endlessly, that you are not enough. That your craft, honed over years, can be reduced to a prompt. That your voice is just training data waiting to be optimised. This fatigue is a sane response to being asked to celebrate a system whose primary effect is to concentrate decision-making in fewer hands. Reframe around the three pillars, and the fog lifts.
The Shillings 745 Trillion Tab
Big Tech wants every computer to be AI-capable. Here is the price tag.
There are 2.1 billion computers in active use worldwide. Only 345 million — roughly 16 percent — can run modern AI workloads. Replacing the remaining 84 percent at blended average costs comes to $5.77 trillion. In Kenya Shillings: 745 trillion. That exceeds the entire annual global IT budget.
The sales pitch conceals a wealth transfer. That $5.77 trillion flows to a handful of silicon manufacturers. An NVIDIA H200 GPU each costs KSh 5.6 million landed in Nairobi. An eight-GPU server: KSh 52 million. Kenya adds 25 percent import duty and 16 percent VAT. The hardware bill alone is a debt trap shaped like progress.
Natural refresh cycles will replace non-AI hardware within years. But the urgency narrative demands developing countries leap before the economics make sense, taking on foreign-denominated debt for servers that depreciate faster than the loans. The result: nations that never own their compute, never own their data, never own the models trained on it. They end up poorer and more indebted in future.
The Sun Beneath Our Feet
While celebrity billionaires pitch solar farms in orbit, a patch of North African desert could power the planet.
The solar energy striking the Sahara each day could satisfy global electricity demand several times over. One square kilometre of panels in Morocco generates five to seven gigawatt-hours daily. Scale to 10,000 square kilometres — the size of Jamaica — and you power every home, every factory, every AI data center on earth.
Meanwhile, tech founders talk about space-based solar beaming power to earth. China has crossed 1,100 gigawatts of installed solar capacity — more than the entire United States — and its solar
exports to Africa rose 176 percent in a single month. Kenya straddles the equator yet imports power while the sun beats down on millions of rooftops that could generate free electricity.
While solar panels and batteries are currently duty exempt under East African Community Customs Management Act, The Kenya’s Finance Act 2025 curiously moved them from zero-rated to standard-rated 16% charged at point of sale ~16% cost embedded in supply chain, raising end price without appearing as a line-item tax. Further, the Finance Bill, 2026 (pending in parliament) proposes moving from them zero-rated to exempt affecting electric bicycles, electric buses, lithium-ion batteries and solar panels - distinguishing self-inflicted national electric disaster from the foreign-orchestrated.
The Desertec (Morroco-EU) project collapsed when photovoltaics (solar panels) became cheap enough for Germany to generate its own power. It was never about helping Africa; it was about Europe securing energy. The same pattern repeats with AI: data centers go where electricity is cheap and regulation weak, not where local populations benefit most. Extractive relationship with foreigners will never save us.
The Walled Garden vs. The Commons
The most consequential divide in AI is between openness and enclosure.
Two distinct models compete. The American: proprietary, rent-seeking, closed. Corporations download the entire internet — copyrighted works included, under a judicial shield of Fair Use — train their models, then charge users perpetually. The Chinese: Open Source, releasing systems freely for anyone to download, modify, and deploy. One concentrates power. The other distributes it.
Data flows from the Global South to Northern data centers, is refined into models, and sold back as a service. Open Source is the only framework giving Kenyan creatives genuine agency: the right to use, study, modify, and share. Everything else is a licensing agreement dressed as a revolution.
For a filmmaker in Nairobi or a designer in Mombasa, the choice between a proprietary model charging per generation and an open-source model running locally is not technical. It is political. The proprietary model extracts autonomy along with money: every query trains the next model; every creative decision is shaped by biases baked into a black box. Open source offers the possibility of retraining on local aesthetics and keeping value within the community. Compound with private data donation to foreign servers and creative surrender to alien big tech – and let’s not get into surveillance capitalism and autonomous weapons among other harmful uses mopped data is uses in.
Kenya’s AI models choice bears consequential and a moment to pause and reflect on absent National AI Policy as philosophical scaffolding to a coherent artificial intelligence aspiration?
The Voice Merchants
Taylor Lorenz broke the news. YouTube channels are being bought — narratives and all.
In 2025, journalist Taylor Lorenz reported that major tech corporations had begun purchasing influential YouTube channels outright. Not the ad slots — the channels themselves, lock, stock, and audience trust. A creator with years of built credibility possesses something no campaign can buy. Buy the channel, buy the credibility, buy the ability to shape what millions believe about AI.
The same dynamic plays out in Kenya through sponsored content that blurs genuine enthusiasm with paid advocacy. When you cannot tell whether the person urging AI adoption believes it or is being paid, the information ecosystem becomes unreliable. Confusion favours the incumbent.
Control of narrative has always preceded control of markets. In radio’s early days, manufacturers created their own programmes. Today, the outright purchase of trusted voices. The danger for Kenyan creatives is not replacement by AI tomorrow. It is that the conversation about what AI should be and whom it should serve is being captured by interests that see creativity as a content pipeline.
The Schizophrenic Regulator
On Monday: ‘AI is too dangerous to leave unregulated.’ On Tuesday: ‘Overregulation will cede leadership to China.’ Same industry, both days.
When public concern rises, executives who warn their own inventions pose existential risk pivot to arguing regulation would cripple innovation. The industry writes the rules it pretends to resist.
China has the most comprehensive AI legislation in the world. The United States has a patchwork of toothless frameworks. The debate was never about safety. It was about who gets to decide.
The oscillation is a power play. By alternating alarm and dismissal, the industry keeps regulators off balance. When you cannot decide whether to fear the technology or trust its stewards, inaction becomes the default — allowing unaccountable deployment. Global South nations, with constrained regulatory capacity, are especially vulnerable to this whiplash.
The Innocent Machine
Artificial intelligence is nothing but switches. The morality lives entirely in the humans who wield it.
A language model has no intentions. It has training data and a loss function. When a chatbot gives dangerous advice, the fault is not the matrix multiplication; it is the decision to deploy without safeguards. When AI displaces a graphic designer, the blame belongs to the business that chose labour arbitrage over human creativity.
The same architecture that helps a farmer diagnose crop disease is, in different hands, a tool to replace illustrators with prompts. The technology did not change. The power relationship did.
There is a subtler dimension. An AI that generates beautiful images but has never tasted food, never felt the harmattan wind, never lost a loved one — can it produce work carrying weight? The machine can mimic. It cannot witness. Kenyan creatives who understand this hold an advantage no model can replicate. Intelligence is the sum total of all basic five senses, add the sixth one. Postures of “Sentient AI” are laughable.
Beyond the Binary
The question is not whether you are for or against AI. It is whether you have been offered a genuine choice.
The binary framing — embrace or resist, booster or Luddite — serves those who benefit from a rushed decision. Kierkegaard wrote that ‘the crowd is untruth.’ Sartre called pretending we have no choice ‘bad faith.’ The ‘normie’ label, which emerged around 2015 as a lexical weapon to designate anyone who follows without questioning, is now deployed to shame those who ask whether this technology serves the many or the few.
Nietzsche’s ‘last man’ avoids challenges and makes everything small. Foucault showed how institutions — schools, prisons, social media — are architectures designed to produce normative behaviour by rewarding conformity. When the AI industry labels its critics Luddites, it is deploying a disciplinary mechanism. The ‘normie’ label was never about normality. It was about power.
The genuine choice is between a future where AI is a tool in the hands of many, accountable to many — and one where it is a weapon in the hands of a few. The fatigue creatives feel — called a Luddite for hesitating, a normie for asking questions — is manufactured. The binary itself is the trap.
The More Things Change
The economists Daron Acemoglu and Simon Johnson, in their book Power and Progress, argue that the textile factories of the early British industrial revolution generated great wealth for a few but did not raise worker incomes for almost a hundred years. Too late for the textile workers who lost good jobs. Too late for the Luddites who ended on the scaffold. The question they pose is not whether technology creates jobs. It is whether the society that deploys that technology shares the gains.
There is a reason the new machinery of 2025 looks so much like the machinery of 1812. What made the Luddites desperate was not the technology but the society that deployed it. John Booth knew this. The machine, he said, ‘might be man’s chief blessing instead of his curse if society were differently constituted.’ Two centuries later, the constitution of society remains the decisive variable. The three forces — money, power, control — are the lens through which every AI announcement, white paper, and government strategy should be read. Who pays? Who rules? Who decides? Answer those, and the barrage of a supposedly heavenly future of AI abundance loses its hold. The future is not written by the technology. It is written by whoever owns it.
The invitation to Kenyan creatives is not to reject AI. It is to interrogate it. For every tool: does this increase my sovereignty or diminish it? Does it distribute power or concentrate it? The machine in the mirror reflects not circuits but the society that built it.
Booth’s secret died with him. But his insight — that the machine is only as good as the society that builds it — remains the most urgent question facing the Kenyan creative today. The normies, it turns out, are not the ones who hesitate to adopt AI. The normies are the ones who adopt without asking who it serves.
-----
Released under: Creative Commons Attribution 4.0 International license
Alex Gakuru - Director Center for Law in Information Technology ("Claw-IT") is a consultant on technology policy, cyber law and governance at the intersection of technology and society.
He can be reached through: gakuru@gmail.com