Kenya had spent more than a decade positioning itself as a digital-work destination. It built infrastructure, trained graduates, and wrote policy around remote labour. Then generative AI arrived at the worst possible moment: the government had just adopted a ten-year national masterplan for outsourcing and gig work in 2022, the same year OpenAI released ChatGPT.

The results are visible on the streets of Nairobi. At its peak early in the decade, researchers estimated at least 40,000 people in the capital were paid to do other people’s homework, mostly essays for cheating students in richer countries. That industry has collapsed to almost nothing. It was, in its way, an unlikely Kenyan success story — one that offers an unnerving preview for white-collar workers around the world.

What the work paid

Teresios Bundi grew up in a farming village and moved to Nairobi in 2011. He was the first in his family to go to university, but his real education came online. A friend suggested a new kind of job: writing term papers for students abroad. His first assignment paid $7 for three hours of work, a two-page essay about the health benefits of a fruit he had never heard of. The client was a stranger. The topic was a starting point for what would become a twelve-year career in what he now calls a parallel education.

Over those years, Bundi estimates he wrote more than 2,500 essays, sometimes three a day, while completing his own degree in public health. Some students were so trusting, or so detached, that they handed over their university logins so he could keep up with their course work. He graduated in 2015 but never used the qualification in the way he had once intended. Instead, he started his own business, charging $40 to $70 per paper.

He rented a house near a Kenyan university and filled it with desks, fast internet, and mattresses. The students he employed wrote in the gaps between their classes and slept when they could not make it home. For them, the work paid far more than public health would have. Around Nairobi, the money was visible in new Subarus, current phones, and tables at the clubs. For a generation with few formal options, essay writing had become a lifeline.

The policy built on the same ground

Cheap high-speed connectivity reached Kenya around 2009. The government quickly saw an opportunity: online outsourcing could lift university graduates into the global economy. Transcription, data entry, graphic design, web development, and basic coding were all marked as possible exports. Essay writing was never officially condoned, but it fit the same shape. It was remote, digital, and powered by young educated Kenyans.

In 2016 the state began training university graduates to work on freelance platforms. In 2022 it adopted the National Digital Masterplan, a ten-year push that leaned further into outsourcing and gig work. The underlying problem was real. More than 100,000 students complete Kenyan universities each year, formal jobs are scarce, youth unemployment remains above 25 percent, and about 40 percent of the country’s 54 million people live in poverty. That shortage of opportunity helped feed the recent Gen Z protest movement.

Government officials still insist the outsourcing strategy can work. John Tanui, principal secretary at Kenya’s ministry for digital economy, has said he stands by the country’s potential as an outsourcing hub and believes it can unlock opportunities for young people. But the timing of ChatGPT’s launch and the new masterplan could not have been worse for the people who had already invested years in this work.

Transcription went first

The essay industry was not the first Kenyan online job to be consumed by automation. That distinction belongs to transcription, which began drying up in 2017, years before ChatGPT became a household name.

Frida Mwangi, a mother of four, started transcribing in 2015. She worked on interviews with Ellis Island immigrants and helped process medical files. Around 2017 she noticed the assignments disappearing. Speech recognition software had advanced enough to handle many of the recordings she used to receive. Employers began asking her to run audio through AI and fix the errors, but the models mangled medical terms badly enough that she found it slower than starting from scratch. The jobs went anyway. “My job went completely,” she said. She now campaigns for gig workers’ rights.

Kenya’s outsourcing boom also faced criticism over pay and working conditions, and part of the business retreated on its own. Sama, which hired Kenyans to moderate content for Meta, cut back operations. Scale AI discontinued some of its Kenyan work. The industry had always promised a smoother path than what many workers experienced.

A number that says this is not finished

Scale AI and the Center for AI Safety jointly run an index that measures how much online freelance work leading models can complete, across categories like product design and data analysis. The trend is startling. Last October, the models managed to complete 2.5 percent of the tasks. By July, that share had jumped to 16 percent. The jump in a few months suggests the pace of disruption is accelerating.

Mark Graham, an Oxford professor who studies the global gig economy, has described the situation plainly. AI has come in and taken big chunks of these labour markets, he said, and Kenya will not be the only place where that happens. The evidence is already visible beyond academic writing.

Amazon has announced that Mechanical Turk, the microtask platform that began paying people a few cents to do things software could not handle, will close on 30 September. The date arrives twenty-one years after the service opened. Meta spent the first half of this year planning to cut some teams by 60 percent and hand the work to AI. Uber, after expanding across Africa, pulled out of Nigeria and Uganda while also cutting a tenth of its workforce.

What is left is called humanising

Essay work has not vanished completely. The jobs that remain are mostly for people who edit AI-written papers so they will not trip the detection software universities now run. One writer, who would only give her first name because she fears for her future employment prospects, says business is as good as ever. She drafts with AI, rewrites by hand, and avoids the tools built to disguise machine text, preferring her own skills. In her experience, students without writing experience are the ones who get caught.

Plagiarism detection is not reliable either. One AI detection system flagged a Berkeley professor for an op-ed she had written herself. The demand side of the equation has shifted too. An MIT committee found in August that AI can answer almost any written assignment in its undergraduate curriculum. That is why a student no longer needs to hire a stranger in Nairobi to complete the work.

The same wave everywhere

Bundi still lives in Kenya, but no longer writes essays. He now works for a German government development organisation, helping young Kenyans find opportunities in the digital economy. He misses the work, the deadlines, and the subjects he had to learn from scratch. He called the essay-writing years a parallel education, one that trained him in research, argument, and speed.

He has also become a reluctant prophet for the wider labour market. “A.I. is coming for bankers, for accountants, it’s coming for engineers, and A.I. will come for architects,” he said. “It’s coming for everybody.”

His own arithmetic is narrower. The peers who stayed in public health after 2015 have risen through the ranks. If he returned to public health today he would start at entry level, working for someone who graduated last year. The same logic applies to the AI transition. The skills people learned in the old economy are not worthless, but they may not count for as much as they once did.

Source:TNW | Future-of-work News