One of the big questions today is: How can higher education institutions reform assessment so that degrees signal genuine human competency, and not prompt-engineering proficiency, especially in the Global South, where graduate unemployment exceeds 60 per cent?
The ivory towers of academia are dismantled, not due to the external attack, but by the internal decay that endangers to make higher education a sophisticated version of credential laundering.
The emergence of such graduates who are able to write excellent research papers but are not able to defend their positions face-to-face, who get Honours degrees but still cannot express even the simplest ideas in a job interview, has become a disturbing agenda across universities in the Global South.
This competency crisis is not due to a lack of ideas but an evaluation system that prefers the ability to use the computer well over the ability to show that one really understands the material.
Large language Models (LLMs) revolution (including ChatGPT, Claude, Gemini and others) has revealed this basic defect in our academic structure. Eighty-nine per cent of students admit to using generative AI for their assignments, thus creating an educational paradox: the more sophisticated our assessment methods become, the less our graduates prove competent in practical applications.
As a result, universities have transformed into assembly lines producing algorithmically enhanced degrees, instead of being places where true intellectual development takes place.
READ MORE: Why Nigeria’s New University Admissions Policy is Backwards and Anti-Human Rights
According to the World Bank, the employment statistics are really bad. In Tanzania, Kenya and Ghana, the ratio of unemployed university graduates three years after graduation is more than 60 per cent, and it is not due to economic stagnation but rather to the fact that they do not possess the required skills by the employers: critical thinking, communicative competency, and the capability to combine knowledge under pressure.
No AI can take over these skills, but still, our evaluation criteria are giving points to the ones who have perfected the craft of prompt-engineered delegation.
Alternative assessments
What alternative assessments would be created if we substituted the conventional written coursework with compulsory oral examinations and module-level publication requirements, thus changing the whole scenario of assessment from a private keyboard exercise to a public voice-verified performance showing one’s understanding?
Researches show that oral exams increase student motivation and lead to the acquisition of concepts, especially for first-generation students who are usually at a disadvantage in traditional settings.
When students are required to express their understanding verbally, they interact with the AI content on a more profound level and acquire the communication skills that various employers have consistently indicated as lacking the most. An oral defence fosters a level of intellectual accountability that is beyond the reach of any LLM.
In addition, requiring each student to publish their work for every module signifies a multi-stage quality control approach, consisting of plagiarism checks, AI detection tests, and most importantly, public peer review.
READ MORE: What Would It Take to Achieve an Innovative Tanzania?
This openness, stakeholders argue, sets up a natural selection system where poor-quality research is immediately rectified, and great work gets the benefit of citations that uplift both the student and the institution’s reputation.
Publication also brings academic isolation that was caused by LLM’s drop-boxes to an end. Students who are aware that their papers will become public are said to see research as taking part in ongoing conversations rather than just a process of getting the grade.
The turn from “submit to professor” to “publish to the world” educates the learners in the direction of real knowledge creation.
Integrity incentives
Think of the integrity incentives. At present, an LLM is able to come up with a reasonable essay in a matter of seconds, while the detection tools are continually behind in the updates of the models.
If students have to support their arguments through oral presentation and their name is attached to a permanent PDF, the cost-benefit calculation changes: the energy that used to be spent in prompt-engineering will now be redirected to authentic comprehension since superficial knowledge will be revealed in real-time questioning.
A new look at assessment is drawn on the presentation-based assessment architecture refined by Professors Komaruddin Hidayat, Farish A. Noor, Associate Professors Djayadi Hanan, Moch Faisal Karim, and Sirojuddin Arif at Universitas Islam Internasional Indonesia (UIII).
READ MORE: Unemployment’s Lure: Why Young People Are Vulnerable to Bad Influences
Their mandatory public seminars and viva-style module closures produce graduates who are simultaneously tech-literate and voice-verified. Likewise, UIII senior lecturers Nia Deliana, Zezen Zaenal Mutaqin, Afrimadona, Samuel Blanch, Rifqi Muna and Ridwan have demonstrated that semester-long publication pipelines (student papers are peer-reviewed, copy-edited and uploaded to an institutional repository) can be scaled efficiently in resource-constrained contexts.
Their data show oral defence cohorts outperform traditional-assessment cohorts on employer satisfaction metrics by 34 per cent. These Indonesian experiments prove the model is neither utopian nor Western-centric.
Skepticism
Critics object that oral exams and publications are too staff-intensive. Yet the hidden costs of the status quo (graduate unemployment, degree devaluation, reputational damage) far exceed the investment required (author’s cost-benefit modelling).
Panels of three faculty can examine twenty students in a single afternoon; open-source journal platforms such as Open Journal Systems (OJS) eliminate print overheads; and teaching assistants gain pedagogical training by serving as editorial reviewers.
The job markets have already made their choice: recruiters are now relying on writing tests and assessment-centre presentations in addition to academic transcripts, as degrees are no longer considered a guarantee of competency.
The universities which will synchronise credentialing with these real-world signals are going to take the lead in global ranking determined by employer reputation.
READ MORE: Disoriented Education: Why a University Degree Is Now More Questionable Than Ever?
The implementation road map is quite simple. Phase 1: An oral defence will be conducted in capstone courses on a pilot basis with random assignment of 50 per cent of the sections, which will help in generating counter-factual evidence.
Phase 2: Publication requirements will be gradually introduced, with first-year annotated bibliographies, second-year literature reviews, and upper-level original articles (where a searchable repository that promotes teaching quality will be built) as the levels of publications.
Phase 3: There will be training for faculty through workshops that are conducted by Indonesian pioneers whose procedures are Creative Commons-licensed and are accessible in both Bahasa Indonesia and English.
Stark choice
The choice is stark. Continue minting keyboard warriors whose eloquence ends at the chat-box, or graduate voice-verified scholars who can think, speak and write without algorithmic scaffolding.
In an era where LLMs can instantly produce plausible prose, only human-centred assessment (oral, public, citable) can preserve the university’s covenant with society.
The LLM has exposed our educational deficits; now, human judgment must restore academic excellence. The Indonesian experience proves it is not only possible but imperative.
READ MORE: It Is High Time Tanzania’s Universities Start Preparing Graduates for Prosperity Instead of Poverty
To the vice-chancellors, deans and curriculum designers of Africa and the Global South: the Indonesian experiment is replicable tomorrow.
As your national visions target a knowledge-based economy, adopting OJS-hosted student journals and mandatory viva-style module closures would cost each campus less than a single subscription to Turnitin’s AI-detection suite while creating an open-access corpus of Global South scholarship that Google Scholar already crawls for free.
Crucially, an A-grade should not only result from class attendance and offline assignment submission, but is most required by submitting the papers online, thereby locking transparency and global citability into every high mark.
Cease importing Western surveillance tools and instead export a home-grown, voice-verified credential that signals to investors, NGOs and continental employers that a Global South degree means a graduate who can think, speak and publish without algorithmic scaffolding.
Peseo Lao Pio is a Tanzanian national currently pursuing a Master of Arts degree in Political Science at the Indonesian International Islamic University in Depok, Indonesia. He’s available at pio.peseo@gmail.com. The opinions expressed here are the writer’s own and do not necessarily reflect those of The Chanzo. If you are interested in publishing in this space, please contact our editors at editor@thechanzo.com.
