If AI Is Taking the Entry-Level Jobs, What Exactly Should Kenyan Schools Be Preparing Students For?

In November 2018, a doctoral candidate who had enrolled at Kenyatta University nine years earlier told a conference hosted by the Commission for University Education (CUE) that he was still waiting to graduate. Another speaker at the same gathering noted that some faculty members were supervising up to 70 master’s and PhD candidates at the same time, despite CUE guidelines capping the load at three doctoral students per supervisor. Writing a doctoral thesis on Yvonne Adhiambo Owuor’s novels Dust and The Dragonfly Sea typically takes three to four years at a British university and slightly longer in the United States. Inside Kenya’s congested postgraduate pipeline, the same degree could drag on for nearly a decade.

Today, a free large language model will generate a thesis-length draft on those two novels in a single afternoon, complete with chapter headings, a literature review, textual analysis, and a partly fabricated bibliography. The grammar holds together, the chapter layout follows convention, and the surface argument reads well enough to slip past an overworked external examiner. Kenyan literary editors are already dealing with the fallout.

At an eKitabu literary festival in May, magazine editor Phyllis Nyambura noted that many young writers were sending in flat, formulaic short stories that had clearly been written by artificial intelligence, according to the Daily Nation. Earlier this year, the British literary journal Granta, which publishes the regional winners of the Commonwealth Short Story Prize, ended its external publishing partnerships after critics claimed that some winning entries read as though they were partly written by machine, an allegation the authors denied.

Consumer internet tools never crossed that line. Even when Google and YouTube gave pupils instant access to both source material and distraction, the student still had to read the pages, weigh what mattered, and write the sentences. Large language models do the drafting themselves. Once software can counterfeit the finished product of an education at almost no cost, the agency responsible for deciding what Kenyan children learn and how they prove they have learned it has to rethink the classroom from scratch. That mandate sits with the Kenya Institute of Curriculum Development (KICD).

Even as Nairobi markets itself to foreign venture capital as the Silicon Savannah, the state bodies running its schools have responded to generative AI with a two-day KICD vetting workshop, a World Bank-funded drive to install smartboards in junior schools, and a circular from the Kenya National Examinations Council (KNEC) instructing teachers not to use chatbots to complete their pupils’ coursework. A review of public filings shows no national framework from either the Ministry of Education or KICD instructing teachers on how to teach or grade a teenager who has a free chatbot waiting on a phone at home.

By treating artificial intelligence as another software package to graft onto the existing timetable, education officials have missed the structural damage. Generative AI has broken the unsupervised assessment model that underpins both the Competency-Based Curriculum (CBC) and the Kenyan university degree.

Why unsupervised homework no longer measures learning.

The mark written in red ink at the top of an essay was always the least important part of setting homework. What actually formed the student was the slow, frustrating work required to finish the page: sitting with an unfamiliar text, piecing together an argument, pushing through boredom, and handing the work in against a fixed deadline. Working through that resistance strengthened working memory and the ability to read long texts without losing the thread. It also built the habit of delivering finished work on time. In 1973, the economist Michael Spence, who later shared the 2001 Nobel Prize in Economics, showed that employers rely on educational credentials to gauge qualities they cannot observe directly during a job interview. When a Nairobi bank or law firm hired a graduate with an upper second-class degree, it inferred that the candidate could sit down alone and work through an unstructured problem.

Chatbots strip out that productive struggle while leaving the finished document intact, and early classroom studies show the cost of taking away the friction. In a 2025 field experiment tracking nearly 1,000 secondary-school mathematics pupils in Turkey, researchers led from the University of Pennsylvania found that students given unrestricted access to GPT-4 scored 48% higher than a control group during practice sessions, only to score 17% lower than the control group once the chatbot was removed for the final exam. By contrast, pupils given a restricted tutor version programmed to offer hints while withholding final answers scored 127% higher during practice and suffered no drop when tested unaided. A separate, pre-publication study posted by the MIT Media Lab in June 2025 found that university students who drafted essays using ChatGPT recorded the lowest neural connectivity across three test groups and struggled to quote from essays they had handed in minutes earlier.

Across Kenyan campuses and secondary schools, students have already moved onto the tools. In December 2025, The EastAfrican reported on a survey by Nairobi research firm Stadi Analytics showing that almost every Kenyan student surveyed used AI tools for daily coursework, with nearly a third admitting that chatbots had "somewhat replaced" actual studying. Whenever a secondary school or university sets a take-home essay that a phone app can finish in half a minute, coursework marks climb while comprehension drops, leaving the school to rubber-stamp a pupil’s ability to copy and paste a prompt.

That flaw cuts straight into the design of the CBC. To move away from high-stakes final exams, the curriculum shifted substantial weight onto continuous school-based assessments that feed directly into a learner’s national grade. In September, KNEC issued guidelines for the 2026 school-based assessments warning that teachers "must not use AI to complete or assist learners" with their assigned projects, while noting that the rule was not a total ban on AI in schools. By focusing on teachers who might falsify marks on behalf of their classes, the council left untouched the far wider habit of pupils generating polished coursework at home on a parent’s smartphone.

Buying school hardware without a classroom method.

KICD’s main public step on artificial intelligence took place in October, when it partnered with the Commonwealth of Learning to host curriculum developers, teacher trainers, and classroom teachers in Nairobi for a two-day review of AI-generated lesson materials, Education News reported on 19 October. Participants checked machine-drafted content for factual accuracy, syllabus alignment, and bias alongside sessions on copyright, open licensing, and student data privacy. KICD Chief Executive Charles Ong’ondo called on teachers to test AI tools while keeping human oversight in charge, speaking shortly after the institute won an African public-sector curriculum award in Accra.

At the Ministry of Education, the serious money is still flowing into physical screens. In July, Deputy President Kithure Kindiki launched a rollout under the World Bank-financed Kenya Digital Economy Acceleration Project to supply 10,382 public junior schools with more than 20,700 devices, combining interactive smartboards with laptops for teachers. Information and Communications Technology Cabinet Secretary William Kabogo said the hardware would prepare pupils for an "increasingly AI-driven world", with officials adding that over 62,000 teachers had completed technology integration training. Launched in March 2025, Kenya’s national AI strategy puts the cost of implementation through 2030 at a minimum of KSh 152 billion, earmarking half of that budget for digital infrastructure. By November 2025, President William Ruto was publicly pressing universities "to integrate AI into their teaching, research, and institutional operations".

Kenyan taxpayers have funded a nationwide school hardware drive once before. Pledged during the 2013 general election campaign and rolled out from 2016, the Digital Literacy Programme distributed roughly 1.2 million tablets across 22,891 primary schools at a cost exceeding KSh 32.2 billion. When reporters from the Daily Nation audited the program after the 2022 election, they found thousands of tablets locked in store cupboards, broken, or stolen, with headteachers pointing to missing syllabus content, unreliable power and internet, and rushed teacher training. Buying hardware without settling how children actually learn turned school storerooms into electronics warehouses.

Detailed international templates already sit on the shelf if KICD chooses to use them. In September 2024, UNESCO released twin AI competency frameworks for school pupils and teachers, mapping out 12 core competencies for learners and 15 for educators that put ethics, critical evaluation, and human judgment ahead of software operation.

Before shipping another wave of screens into primary classrooms, education officials in Jogoo House should look closely at the countries now pulling them out. After the Karolinska Institute warned in 2023 that "there’s clear scientific evidence that digital tools impair rather than enhance student learning", Sweden’s government scrapped its mandatory digital-schooling plan, put SEK 685 million into buying printed textbooks in 2023, kept funding paper books thereafter, moved to end mandatory screen access in preschools, and kept Year 3 national literacy and numeracy tests on paper. Across the border in Norway, Prime Minister Jonas Gahr Støre told public broadcaster NRK in March that new national rules taking effect on 1 August would strictly curtail screen use across the first four years of primary school. Meanwhile, Australian states have barred mobile phones from state schools and, since December 2025, federal law has prohibited children under 16 from holding social-media accounts. Kenya’s 2016 tablet drive aimed straight at six-year-olds in Standard One, the exact age group Scandinavian ministries are now pulling away from screens. None of KICD’s published briefs weighs that medical and cognitive evidence.

Inside Kenyan classrooms, the daily routine remains anchored in chalk-and-talk dictation followed by memory tests. For generations, secondary school chemistry classes have drilled acid-base titration to match a KNEC marking scheme without explaining how a technician actually uses titration on the floor of a brewery, a drug factory, or a municipal water plant. Although the CBC was meant to replace rote memorization with practical competence, the transition is buckling under administrative strain. On 20 January 2026, Education Cabinet Secretary Julius Ogamba disclosed that just 75% of the 1,130,459 learners who sat the junior school assessment had reported to Grade 10, leaving roughly 280,000 teenagers unaccounted for at the senior school gates. Dropping chatbots into an overstretched school system built on copying notes off a blackboard simply automates the copying.

Running universities on ten-year-old notes and KSh 98 billion in debt.

Students leaving Grade 12 walk into public universities that are already technically insolvent. By March, accumulated debts across Kenya’s public universities had climbed to roughly KSh 98 billion, The Standard reported, prompting members of the National Assembly’s education committee to warn that several campuses faced financial collapse.

Two years earlier, the chief executive of the Universities Fund told the same parliamentary committee that public universities were holding onto roughly KSh 24.6 billion in unremitted staff pension deductions alongside KSh 20 billion in pay-as-you-earn income tax deducted from lecturers’ payslips but never wired to the taxman. Meanwhile, the government’s student funding formula, challenged in court and struck down as unconstitutional by the High Court in December 2024, kept alive under a Court of Appeal stay in March 2025 and rewritten again this year remains mired in dispute.

In the 2026/27 budget estimates, higher education receives KSh 163.9 billion, over half of which goes directly to student loans and scholarships, alongside KSh 58.5 billion for technical and vocational colleges, while the KSh 98 billion arrears pile was left outside the ministry’s budget ceilings, The Standard noted in February. On the campaign trail, the administration has poured its energy into technical training institutes and bilateral deals to send Kenyan tradespeople, nurses, and drivers to work overseas. Every growing economy needs skilled electricians, plumbers, and clinical officers. Allowing research universities to rot at the exact moment artificial intelligence reshapes the global economy, however, strips the country of the legal, mathematical, and scientific capability required to build, test, and regulate intelligent systems of its own.

Academic standards inside the lecture theatre were fraying long before ChatGPT arrived. Undergraduates across Kenyan campuses refer to dog-eared lectures read aloud unchanged for a decade as "yellow notes", taught by part-time adjuncts who shuttle between three or four universities each week just to piece together a living wage. When a university unit relies on ten-year-old dictated notes and a generic take-home term paper, a free chatbot makes the assessment obsolete overnight. "The biggest mistake would be to assume that because an answer is well written, it is necessarily correct," Moses Gitonga, a part-time lecturer at Kenyatta University, told Capital FM in August. A 2025 survey covering 137 faculty members across 13 Kenyan universities found lecturers split between excitement over AI’s research speed and alarm over student plagiarism, data leaks, hallucinated citations, and cognitive laziness.

At the same time, the entry-level job market that justified the price of a degree is narrowing. For decades, a Kenyan bachelor’s degree worked as a screening ticket into formal white-collar employment, where fresh graduates spent their first two years drafting basic legal contracts, reconciling spreadsheets, summarizing research briefs, writing boilerplate code, or turning out marketing copy. Generative AI now handles those exact first-rung tasks in seconds. In an August 2025 working paper, Stanford economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen found that US workers aged 22 to 25 in the white-collar roles most exposed to generative AI suffered a 13% relative drop in employment after the tools went mainstream, while the World Economic Forum’s employer survey projects that 39% of core workplace skills will shift by 2030. When corporate graduate intakes shrink at the same time as universities hand out degrees based on AI-written coursework, the value of a Kenyan degree gets squeezed from both ends.

Foreign universities have already started tearing up their exam rules to separate what a graduate knows from what software can generate. Ahead of tighter Australian higher-education rules taking effect in 2026, the University of Sydney divided every 2025 course assessment into two distinct tracks: supervised, in-person assessments that verify what an individual student can do unaided, and open assessments explicitly designed to test how well students direct and critique AI tools. Kenya’s university regulator has taken an early step in the same direction. In August, backed by funding from Germany’s DAAD academic exchange service, the CUE brought vice-chancellors and academics together for a three-day workshop to draft a national AI framework for higher education. Its working principles call for grading methods that verify real student competence rather than relying on faulty AI-detection software or blanket bans, while warning that academic performance should not depend on whether an undergraduate can afford a $20 monthly subscription to a frontier model. Eight weeks on, that framework remains in draft form.

Avoiding a new digital dependency.

When European powers met in Berlin in 1884 and 1885 to draw borders across Africa without an African leader in the room, their leverage rested on industrial manufacturing and repeating rifles. In the twenty-first century, the gap runs through compute clusters and proprietary foundation models. A state that owns no sovereign compute, holds no equity in frontier models, leaves its security services unprepared for autonomous warfare, and schools its graduates to accept whatever a foreign chatbot autocompletes locks itself into buying imported intelligence while supplying cheap data labour at the bottom of the stack. Nairobi has already seen what the bottom of that stack pays: in 2023, Time magazine documented that moderators in Nairobi hired through outsourcing contractor Sama to screen toxic text and imagery for OpenAI were earning less than $2 an hour.

Whether Kenya climbs out of that low-wage tier starts in the classroom. Although the government’s March 2025 AI strategy directs half of its KSh 152 billion target budget toward data centres and fibre, server racks can always be imported or rented. Training engineers, lawyers, doctors, and civil servants capable of auditing and challenging an algorithmic model takes fifteen years of rigorous schooling, something a system that rewards prompt-pasting cannot produce.

Left to its usual habits, the state will respond by setting up a new regulatory authority or taskforce whose board members immediately move to license practitioners, collect annual registration levies, threaten fines, and print a glossy five-year strategic plan.

Announcing a committee costs nothing politically and leaves a Tuesday afternoon lesson in a Grade 8 classroom untouched. The only measure that counts is whether the Ministry of Education changes how Kenyan children are taught to read and think, and how KNEC and the universities verify that they have done the work themselves.

Rebuilding how Kenyan schools teach and test.

Curriculum planners at KICD, examiners at KNEC, and academic senates across the universities face a choice among three distinct classroom models.

Five practical shifts would move Kenyan classrooms into that third column.

• Keep Grades 1 to 6 on paper and ink. Following the empirical reversal in Sweden and Norway, primary schools should build foundational literacy and numeracy through printed books, handwriting, and mental arithmetic before putting screens in front of young children. Structured AI tools belong in senior school, once a teenager can already read deeply and spot a flawed argument. Any future ministry tender to buy tablets or laptops for primary pupils should require a published scientific review of learning outcomes first.
• Bring writing and coding back inside the classroom. Because an unsupervised take-home essay no longer proves who wrote it, students should do their reading and AI-assisted brainstorming at home, then write their essays, solve equations, and write code inside the classroom on paper or on offline networks. KNEC should amend its 2026 CBC rules to require that all graded school-based assessment projects be either completed under classroom supervision or defended live in front of the teacher.
• Shift marks to oral defences and live cross-examination. While a free chatbot can turn out a polished five-page chemistry or literature project in sixty seconds, a pupil who outsourced the thinking will fold within five minutes of live questioning from a teacher and three classmates asking why a chemical reaction behaves as it does or why a character in a novel makes a specific choice.
• Configure school AI tools to withhold answers. Instead of teaching titration as a dead mark scheme recipe, science lessons should connect it to quality control in a Mombasa water plant or a Ruaraka brewery, while literature classes put complete East African novels in front of learners alongside critical essays. When senior school students use AI tutors, the software should be configured on the Turkish experiment’s model, programmed to ask guiding questions and offer hints while refusing to hand over the finished answer.
• Publish the CUE university framework and pool access to frontier models. The Commission for University Education should publish its August draft framework, strip ten-year-old "yellow notes" out of accredited degree programs, train lecturers to design invigilated and oral assessments, and negotiate bulk institutional licences so Kenyan researchers and undergraduates do not have to pay out of pocket for frontier AI tools. University senate rules should shift degree weighting toward wet-lab practicals, verified field research, and live viva examinations that no model can sit on a student’s behalf.

What a Kenyan degree will be worth

Every chief executive in Nairobi already knows the cost of hiring a university graduate whose transcript lists straight As and whose first month at the desk reveals they cannot draft a clear two-page memo without help. Corporate retraining has operated for years as an unbooked tax on Kenyan business. Leaving generative AI unmanaged inside schools and universities will lock that skills gap in place, turning out graduates with cleaner coursework folders and weaker analytical habits. Used deliberately, the shock of chatbots gives the Ministry of Education the leverage it needs to finish the reform the CBC started and clear nineteenth-century rote dictation out of Kenyan classrooms.

Four institutions hold the pen: KICD sets the syllabus, KNEC polices the school exams, the CUE accredits university degrees, and the National Treasury decides whether public universities are funded to teach or left to drown in KSh 98 billion of debt. So far, not one of the four has published an assessment rulebook built for a world where every student has a free ghostwriter in their pocket.

The Kenyatta University doctoral candidate who waited nearly ten years to graduate in 2018 at least handed in a dissertation that no software could have written for him. A candidate enrolling this semester can generate a passable 300-page draft before dinner. Whether the examiners sitting across the table know how to interrogate the mind behind the manuscript will decide whether a Kenyan degree still means anything at all.