Artificial Intelligence Should Shorten Degrees at (African) Universities and Make Them Cheaper

Introduction:
African higher education is caught in a bind. Demand for university places is growing faster than the continent can fund them, tertiary enrolment sits far below the global average, and even students who win admission often cannot afford to see a degree through to the end. I argue in this paper that artificial intelligence offers a way out of this bind — not simply by making existing degrees better, but by making them shorter.

My central claim is a provocative one: if AI can genuinely speed up how fast students master material, then universities should shorten standard degree lengths — for instance, compressing a three-year Ugandan bachelor's degree to two years, or a four-year Kenyan degree to three — without lowering the bar for what students must learn.

Why does this matter so much?
Because the true cost of a degree is not just tuition. It is tuition plus housing, food, transport, and the income a student forgoes while studying. Shaving a year off a degree cuts roughly a third of the total cost, while also sending graduates into the job market a year sooner. For students from poor households - the very people open and distance learning (ODL) is meant to serve - that combination can be the difference between finishing a degree and dropping out for lack of fees, a well-documented problem among African distance learners.

Why AI and Distance Learning Are Natural Partners

Open and distance learning already exists in Africa as a way to reach students who cannot access a residential campus — rural learners, working adults, in-service teachers, women facing safety or cultural barriers to leaving home. But it has always struggled to personalize instruction, since tutor-to-student ratios stay high even when teaching happens remotely.

I contend that AI changes this equation in several ways:

Personalized tutoring at scale
Generative AI can adapt explanations and practice questions to what an individual student already understands, effectively giving every distance learner a tireless one-on-one tutor.

Predicting dropout risk
The UK's Open University uses a machine-learning system called OU Analyse that flags, with high accuracy, which students are at risk of failing an upcoming assignment. When tutors act on those alerts, student success rates rise — and the effect is even larger for historically disadvantaged groups.

Cheaper translation and materials production

AI tools are already being used to translate study materials into multiple languages faster and more cheaply, a major barrier in Africa's multilingual university systems.

Lighter administrative load
AI-assisted marking and proctoring reduce the staffing bottlenecks that limit how many students an institution can enrol - a real constraint given the continent's teacher shortages.

Taken together, I argue these shifts do not just make distance learning marginally better. They change its underlying economics by cutting the labour cost of personalization, feedback, and administration. These functions used to require either a large residential faculty or a slow, print-based correspondence model.

The Evidence that AI compresses time
I do not ask universities to take this on faith. AI has a track record of cutting task time in other knowledge-based fields: software developers using GitHub Copilot completed a coding task over 50% faster in one Microsoft-affiliated trial; professional writers using ChatGPT finished tasks over a third faster while producing higher-quality output; and an Anthropic analysis found that curriculum-planning work that normally takes a teacher several hours, could be done in minutes with Claude's help.

I am careful to note that this is not universal — one 2025 study found experienced open-source developers actually worked slower with certain AI tools, despite feeling faster. My lesson from this evidence is not that AI speeds up everything, but that any move to shorten degrees needs to be grounded in evidence about which specific learning activities — mainly structured, feedback-heavy formative work — actually benefit from AI assistance.

This is not entirely uncharted territory. Western Governors University in the US lets students graduate as soon as they demonstrate mastery, regardless of semester length. Arizona State University runs accelerated pathways, letting strong students begin graduate coursework while still undergraduates. I read both of these as evidence that decoupling "time enrolled" from "credential earned" is administratively possible within existing accreditation systems — AI simply makes it safer to extend that model to a broader student population, not only the most self-directed learners.

Why African Universities Have Been Slow to Move

Despite pockets of progress — the University of KwaZulu-Natal's generative AI guidelines, Makerere University's AI and data science centre, Kenya's national push (championed by President Ruto) to embed AI across university teaching — I observe that most African institutions have been cautious. Some of that caution is structural: quality-assurance systems built around credit-hours and seat-time make it hard for a single university to shorten a degree without risking accreditation problems. Some is resource-driven: AI tools remain more common at well-funded universities than historically under-resourced ones. And some is simple faculty pushback, echoing concerns raised even at ASU, where AI-generated course tools have drawn criticism from professors worried about job security and the integrity of their original teaching materials.

My Recommendations

• Rather than pushing for immediate, system-wide change, I propose cautious, evaluable pilots:
• Regulatory "sandboxes" permitting AI-supported competency-based progression in select programmes.
• A controlled pilot compressing one degree programme by a year at a single willing university, benchmarked against a matched cohort on the normal track.
• A regional research consortium to pool evidence across countries.
• Means-tested scholarships that pass the cost savings directly to disadvantaged students rather than letting institutions pocket the efficiency gains.
• Joint quality-assurance standards that any compressed programme must meet before permanent approval.

Conclusion
My wager in this paper is that AI's efficiency gains, if handled responsibly, are a policy lever for equity — not just a tool for making existing degrees run more smoothly. Whether African higher education systems are willing to test that wager, through carefully governed pilots rather than either blanket resistance or uncritical adoption, is now a live policy question.

Them