11-09-26

East Africa needs AI training that employers can trust

A student and lecturer review a business assignment

A student and lecturer review a business assignment

The East African Community has approved a foundational AI module for undergraduates across disciplines, from accounting to education. Announced on 31 October 2026 after a ministerial meeting in Kampala, the decision asks Partner States to consider adoption across degree programmes and gives the Inter-University Council for East Africa (IUCEA) responsibility for a regional implementation framework.

For a business hiring its next graduate, the useful question is what passing that course will mean. An AI-generated purchasing recommendation can look convincing before anyone checks the figures beneath it. Teaching a student to find the mistake would give an employer a reason to take the course seriously.

IUCEA's draft module already addresses part of this problem. In its account of an October validation workshop, the council described a module covering critical assessment of AI-generated information, data protection and responsible use. Universities taking it up will have to turn those aims into work that can be examined.

Show how the decision was made

Take a fictional assignment for business students. A distributor in Kampala needs to decide what to order next month. Students receive sales and stock records containing a few deliberate errors, alongside an AI-generated purchasing recommendation.

The assignment asks them to decide whether the recommendation deserves approval. A student who notices that stock returned by customers has been counted as new sales has a reason to question the suggested order. They may need to correct the records or ask for information before committing the company's money.

The lecturer can then mark the reasoning behind the decision. A short discussion afterwards would help establish whether the student understands why the answer changed. The final report matters, but so does the student's ability to account for it.

There is room here for professional differences. An education student checking a generated lesson needs to consider the pupils who will use it. A business student recommending a purchase needs to understand the consequences for stock and cash. A common foundation can accommodate both.

UNESCO's 2024 AI competency framework for students offers a useful reference. Although designed for school curricula, it places critical judgement and ethical responsibility alongside technical understanding. Universities can draw on those principles while setting assignments appropriate to a degree.

Good assessment has a cost

Reading a student's working and questioning a doubtful conclusion takes lecturer time. Teaching staff will also need preparation and access to the tools they are expected to assess. A university budgeting for the module has to allow for that work.

Access for students needs the same attention. If an assignment depends on a paid account, the institution must decide who covers the bill. If students are expected to practise outside class, it must establish how those without suitable devices or reliable connections will participate.

Otherwise, the assessment risks rewarding access to resources as much as understanding. An expensive subscription should not quietly become an entry requirement for a supposedly common course.

Shared teaching materials could spare each university the expense of starting from scratch. Sample assignments and marking guides would also help lecturers agree what counts as satisfactory work. Institutions would still need to budget for supervised practice and the time spent reviewing it.

IUCEA's workshop envisaged pilots at institutions within the regional AI network that demonstrate readiness. It said these would inform wider adoption by examining delivery, institutional requirements and lecturer preparedness, among other questions.

Give employers something to examine

The proposed pilots offer a chance to find out how much support the course needs before more institutions commit to it. They can also show whether different lecturers reach broadly similar judgements about the same piece of work.

Employers have a practical contribution to make. A firm in Kampala, Nairobi or Kigali could help a nearby university develop an exercise around a familiar workplace decision, using fictional records or material cleared for teaching. That would give lecturers a credible problem without exposing customer information.

Universities should retain control of academic standards. A curriculum tied too closely to one company's software or recruitment preferences would serve graduates poorly when they move to another workplace, where the tools and procedures may be different.

For employers, a clear description of the assessment would make an AI course easier to interpret. It would help distinguish a course that requires students to defend their work from one that records attendance at a series of demonstrations.

Businesses considering partnerships should examine the learning outcomes, assessment guidance and teaching arrangements before attaching their names to promises about employability. Clear requirements would also help smaller firms decide whether to contribute staff time to the course.

A published sample assignment, its marking criteria and examples of how students handled a flawed answer would be a convincing place to start. Employers could then judge the standard for themselves.