Case Studies

The Kenyan Essay Industry Just Collapsed — AI Killed High-Skill Outsourcing First

A New York Times investigation reveals how ChatGPT wiped out a decade-old Kenyan essay-writing industry almost overnight, offering one of the clearest early signals of AI displacing global white-collar labor.

Font size

Kenyan essay writer.

Searching that phrase in Google surfaces a hidden supply chain: freelance writers in Nairobi picking up jobs on Upwork, EssayShark, and similar gray-market platforms, ghostwriting essays, book reports, and admission essays for American college students. Per-piece rates run $15–50. Hourly, these writers earn more than local white-collar workers.

A New York Times report on September 5 pulled this supply chain into the open: once ChatGPT went mainstream, American students almost entirely stopped outsourcing. Orders cratered, and writer chat groups filled with complaints.[1]

This story matters far more than it first appears.

How the essay-writing industry grew

Kenya becoming a hub for American essay writing wasn't accidental. Several conditions stacked up:

  • Time zones. Kenya runs 8 hours ahead of the U.S. East Coast. A student submits at 11 p.m.; the writer delivers at 7 a.m. local time. Right on deadline.
  • English proficiency. A former British colony, Kenya delivers higher education entirely in English. Local university graduates write at a level well above Filipino or Indian writers at the same price point.
  • Low cost. A fresh Kenyan graduate earns around $200–400 per month locally. A five-page essay pays $30, which translates to 3–5x local white-collar hourly rates.
  • Persistent demand. Surveys from international academic integrity organizations, cited in the NYT piece, show 10–15% of American college students admit to using ghostwriters. A stable pool of gray-market demand.

The business ran for years. One writer interviewed by the NYT had been at it since 2018, using the income to buy a house and put his kids through school. Before 2023, he said, he consistently handled 20–30 essays a month.

How AI cut the chain

This wasn't gradual displacement. It was a cliff.

When ChatGPT launched in late 2022, the writers were still laughing it off — "you can spot machine-written stuff a mile away." Then GPT-4o arrived in 2024, followed by GPT-5 and Claude 4.5 in 2025. The joke stopped landing.

Three dimensions collapsed simultaneously:

  1. Price. ChatGPT Plus costs $20 per month, unlimited. A single ghostwritten essay runs $30, one-time. The math is obvious.
  2. Speed. A human writer needs 6–8 hours minimum, plus back-and-forth and revisions. ChatGPT produces a first draft in 30 seconds and iterates indefinitely.
  3. Detection risk. This one matters most. Ghostwritten work is "human-written," and tools like Turnitin and GPTZero are optimized for human-written text. AI-generated content is actually harder to flag — it lacks the "outsourced style" fingerprint. Students are betting the professor won't notice.

The NYT piece profiles a middleman — a company that distributes orders to Kenyan writers — that was still expanding its team in 2024, then cut 70% of its writers in early 2025. Those remaining handle less than one-fifth of peak volume.

Once the orders dried up, the skills stopped being sellable. Kenyan writers can produce English essays but cannot build a local app or draft a local marketing plan. The craft carries zero transfer value in the AI era.

High-skill outsourcing died first

This contradicts the prevailing narrative.

For the past decade, AI displacement stories have focused on blue-collar work: autonomous trucks replacing drivers, robots replacing warehouse pickers. The default assumption was that creative work is safe — "creativity is humanity's last bastion."

The Kenyan essay industry exposes that assumption as wrong.

"English essay writing" is widely considered a high-cognition task: parsing the prompt, reviewing literature, structuring arguments, mimicking academic register. Before GPT-5, full automation was thought to be at least a decade away.

Instead, it became one of the fastest-automated occupations. The reasons are straightforward:

  • Massive training data. Academic papers, textbooks, Wikipedia — LLMs train on this material more thoroughly than almost anything else.
  • Highly templated output. Undergraduate essays follow formulaic structures: introduction, literature review, methodology, conclusion. LLMs excel at formulaic output.
  • Fuzzy evaluation criteria. Professors grade on "does this look like an essay," not "is this correct." On the "looks like" axis, LLMs outperform most human writers.

The takeaway: "creativity" is an overestimated moat. Real moats sit at the intersection of creativity and the physical world, or creativity and human relationships. Pure text, pure cognition — LLMs can muscle in on almost all of it.

The subtext the NYT didn't spell out

The report covers Kenya, but the pattern transfers cleanly to India, the Philippines, Pakistan, Bangladesh — every developing country that took on English-language outsourcing.

Young workers in these countries have spent the past two decades climbing the global value chain through two main channels:

  1. IT services outsourcing — coding, customer support, data labeling.
  2. English content outsourcing — copywriting, essay writing, SEO content.

IT services outsourcing remains stable for now; code requires debugging, context, and client meetings. English content outsourcing, however, is collapsing much faster than expected.

Kenya is the first clear sample. Why:

  • Small in scale, making the whole picture visible.
  • Heavy dependence on a single category (essay writing).
  • High sensitivity to dollar-denominated income at the local economic level.

English SEO content shops in Bangalore and English email-support operations in Manila follow the same logic — different industries, same mechanism. These workers will feel the shock before truck drivers do. They just won't get noticed, because they're outside the U.S.

The real point worth highlighting isn't "AI killed another industry." It's that AI has drawn a clean economic impact curve for an entire industry in a single country — a level of granularity that is extremely rare in AI displacement debates.

Most discussion still operates at "AI will change work" — a granularity so coarse it's meaningless. The NYT piece demonstrates what actually happens: when a billion-dollar industry shrinks 80% in 18 months, the impact lands on specific people, specific families, specific mortgages.

Editor's note

This story deserves a closer look because of its granularity.

The past two years have produced a flood of "AI displacement" takes — "AI will replace 30% of jobs," "AI is the new electricity" — all macro slogans with no concrete mechanism. The NYT did the hard thing: it traced the displacement effect down to one city, one group of people, one craft.

My read: the impact on developing countries will be far larger than on developed ones, and the media coverage will be far smaller. The unemployed don't sit inside the audience of English-language mainstream media. A Kenyan writer defaulting on a mortgage or an Indian SEO writer losing a contract won't make headlines, but it will shape millions of lives.

How does this differ from past automation? History includes a few rounds of "white-collar displacement" debate — 1990s ATMs replacing bank tellers, 2010s RPA replacing data entry. Those were either narrow job categories or slow transitions. The Kenyan essay collapse is a cliff — an industry that went from boom to bust in 18 months with no buffer in between.

Caveat: I don't have official Kenyan data on the essay-writing industry's GDP contribution, and the NYT piece doesn't provide specific figures either. The "billion-dollar industry" and "80% shrinkage" numbers are estimates drawn from interview subjects and may not be precise. But even at half that scale, the shock is still remarkable.


References

[1] New York Times, The End of the Kenyan Essay, 2026-09-05. https://www.nytimes.com/2026/09/05/technology/kenya-college-essays-ai.html

Hacker News discussion: https://news.ycombinator.com/item?id=49613642 (24 points / 6 comments, accessed 2026-09-06)

Source: https://www.nytimes.com/2026/09/05/technology/kenya-college-essays-ai.html