Generative AI did not merely automate a task; it destroyed the price advantage of a labour market that once supported tens of thousands of educated Nairobi workers.
Kenya’s collapsing online essay-writing industry is an uncomfortable but useful case study in what artificial intelligence can do to a labour market. The activity itself helped overseas students cheat, so its decline is difficult to defend ethically. Yet the economic mechanism matters because the workers were real, the incomes were real and the technological substitution happened with unusual speed.
The New York Times reported that at the industry’s peak at least 40,000 people in Nairobi were paid to write academic assignments for students overseas. Many were university graduates who could earn considerably more from ghostwriting than from formal employment in their trained professions. One former operator, Teresios Bundi, said he charged $40 to $70 per paper and earned multiples of what a public-health career would have paid.
Then generative AI changed the customer calculation. A student who previously paid a Kenyan writer could obtain a draft instantly from a chatbot at negligible marginal cost. Orders declined, rates fell and businesses that had employed teams of writers disappeared. Entrepreneur and The Next Web have both highlighted the episode as an early warning for digital labour markets in the Global South.
The important mechanism is not simply automation. It is price compression. Kenya’s essay writers had competed internationally because they offered skilled English-language labour at a price attractive to Western customers. AI did not have to become a perfect writer to destroy that advantage. It only needed to become good enough that the customer no longer saw the human premium as worth paying.
That distinction applies far beyond essay writing. Many digital gig jobs are exposed where the buyer values an output rather than a relationship: transcription, basic copywriting, image editing, simple coding, translation, customer support, data processing and some research tasks. If AI can produce an acceptable first version instantly, the value of human labour shifts from production to verification, judgement, integration or accountability.
Kenya’s experience is especially important because the country actively promoted online outsourcing as an employment strategy. Cheap connectivity and a large English-speaking graduate population created a labour pool capable of serving global platforms. In 2022, the year ChatGPT was launched, Kenya’s national digital strategy continued to emphasise outsourcing and online work. The policy assumption was that connectivity could connect surplus educated labour to global demand.
AI changes that equation because connectivity now connects both workers and software to the same customer. The worker is no longer competing only with another freelancer in India, the Philippines or Eastern Europe. The worker is competing with a model whose marginal cost approaches zero.
This does not mean digital work disappears. It means the durable jobs move up the value chain. Remaining essay work reportedly includes “humanising” AI-generated text to evade detection. Other sectors will develop similar hybrid roles. But jobs built solely around producing standardised digital outputs will be under persistent pricing pressure.
For African policymakers, the lesson is to separate digital employment from digital capability. Training people to obtain tasks on global platforms can create income, but it does not automatically create resilience. Workers need skills that remain valuable when tools improve: domain expertise, client management, complex problem solving, project ownership, sales, compliance and the ability to use AI to increase output rather than compete against it.
For companies, AI adoption should also be understood as labour-market restructuring rather than only productivity software. A business that replaces entry-level tasks with automation may save costs today but also remove the apprenticeship layer through which future specialists learn. Organisations will need new ways to train people if junior production work disappears.
The Kenya case also raises the question of income transition. Workers who spent years in a high-paying informal digital niche can struggle to re-enter formal professions because their experience is difficult to certify. Labour policy, professional training and entrepreneurship programmes need to recognise those transition costs.
The employment risk is particularly sharp for countries where formal job creation already lags behind the number of graduates entering the labour market. Gig platforms absorbed people who were educated but underemployed, often without requiring employers to make long-term commitments. AI can remove those opportunities faster than formal sectors can create substitutes. That raises a policy question about transition speed. Education systems typically change over years, while software capability can change within months. Governments, universities and employers therefore need shorter feedback loops between labour-market signals and training programmes. The objective should be to help workers move into roles where AI is a productivity tool rather than a direct replacement.
The decisive point is not whether academic ghostwriting deserved protection. It did not. The point is that AI erased the economic advantage of a sophisticated online labour market almost overnight. African economies that see digital outsourcing as a mass-employment strategy should treat Kenya’s experience as an early signal: connectivity creates access to global work, but only capability that moves faster than automation creates durable bargaining power.
