Waiting for Makola

On July 2, Prof. Isaac Wiafe of the University of Ghana's Department of Computer Science delivered his inaugural lecture at the Great Hall under a deliberately provocative title: “Why AI Is Irrelevant to Ghana: Reclaiming Our Future Through Human-Centred Transformation.” The provocation worked. GhanaWeb ran it as “UG Prof explains why AI is irrelevant to Ghana,” and the headline is now doing its rounds through WhatsApp groups and newsrooms, shedding nuance at every forward.

I want to engage the lecture seriously, because Prof. Wiafe deserves better than his headline — and because the parts of his argument that are wrong are wrong in ways that matter for national policy.

Let me begin with the concessions, because they are substantial.

Wiafe's central warning is that the greatest danger facing Ghana is not failing to adopt AI, but adopting it extensively while failing to develop the capacity to create, govern, and capture value from it. Readers of this site will recognize that argument: it is the subscriber-versus-stakeholder problem I laid out in Who Owns the Meter?, and the professor states it with precision. Access is not ownership. Consumption is not capability. On this, he and I are allies.

His first prescription is equally sound. He calls the building of Ghanaian data infrastructure — high-quality datasets in local languages for education, health, and agriculture — the most urgent task before us, and names the absence of local content as one of the most pressing barriers to making AI serve this country. I made the same argument in Data Before Infrastructure: undigitized archives train nobody's models, least of all our own. That his lab at Legon has already begun this work, with support from partners like Google, is exactly the kind of quiet institution-building this country needs more of.

So this is not a rebuttal of the professor's diagnosis. It is a rebuttal of his sequencing, one buried assumption, and the political afterlife of his headline.

The load-bearing word in Wiafe's lecture is until. AI is not relevant to Ghana, he argues, until district assemblies use it to solve local problems, until traders in Makola and Kejetia use it to improve their businesses, until farmers produce more with less uncertainty. The picture is right. The tense is wrong.

Relevance is not a precondition that a technology must satisfy before a nation engages with it. Relevance is what adoption produces. We have run this experiment before, on this soil, within living memory. Mobile money was irrelevant to the market woman in Makola — right up until the day it wasn't. Nobody waited for the service to prove its relevance to her before building it; she made it relevant by using it, imperfectly, for purposes its designers never listed. Today sub-Saharan Africa moves roughly two-thirds of the world's mobile money value. Had we demanded demonstrated relevance to the ordinary trader as the entry condition, we would still be demanding it.

The same is true now. The trader in Makola does not need a Twi-language frontier model to begin. She can use the tools that exist today — in English, in Pidgin, through a voice note, through her daughter's phone — to track inventory, draft a supplier message, compare prices, understand a bank form. Crude, partial, unlocalized: yes. Irrelevant: no. And every one of those crude uses generates precisely the local demand signal, the usage data, and the popular fluency that Wiafe's own data-infrastructure project requires to succeed. Waiting for relevance and building relevance are not the same activity. One of them is a schedule for never starting.

And here the professor's premise collides with the ground truth. The transformation he says has not yet begun is already underway — quietly, unevenly, and largely unrecorded. Walk through his own list.

Traders. The women of Makola and Kejetia use AI every day, mostly without knowing it. Machine-learning fraud detection guards every mobile money transaction they make. Recommendation and advertising algorithms already drive their sales on WhatsApp, Instagram, and TikTok. Translation tools bridge their supplier conversations. AI-assisted bookkeeping apps sit in the same phones that hold their MoMo wallets.

Schools and clinics. Students are using AI tutoring and exam preparation tonight, with or without ministerial approval. Teachers are drafting lesson plans with it. And in Accra, minoHealth AI Labs — a Ghanaian company — has been building AI-assisted medical imaging for African hospitals for years, which happens to satisfy the professor's own test: Ghanaians creating and capturing value, not merely consuming it.

Persons with disabilities — the group Wiafe rightly centers — may be the clearest case of all. Speech-to-text, text-to-speech, real-time translation, and image recognition for the visually impaired are on ordinary smartphones today, delivering more practical inclusion this year than a decade of policy documents.

And the ordinary Ghanaian with nothing more than a smartphone is using AI constantly — through search, voice assistants, predictive text, spam filtering, navigation, photo enhancement, and, increasingly, the chatbots themselves.

So the challenge is not that AI has yet to arrive in Ghana. It has. The challenge is accelerating adoption, expanding digital literacy, building local talent, and capturing more of the value chain. Those are two very different arguments — and framing AI as something not yet relevant risks overlooking the quiet revolution already taking place in the hands of millions of Ghanaians every single day.

Wiafe argues that Ghana may not need the largest frontier model in the world — that we can compete strategically in efficient training, low-resource methods, domain adaptation, and the evaluation of local content. As a research agenda for a university department, this is exactly right. As a national posture, it conceals a trap.

The trap is reading “we don't need the frontier model” as “we can sit out the utility while we prepare our own.” Local datasets and low-resource methods are complements to the global intelligence utility, not substitutes for it. The farmer does not care whether the model that diagnoses his crop disease was trained in Accra or San Francisco; he cares that it works this season. A posture that defers deployment until the Ghanaian layer is ready inverts the dependency: the Ghanaian layer will be built fastest by a population already fluent in using these tools, already generating local data, already surfacing the failure cases that tell researchers where domain adaptation is needed.

Then there is the sentence that gives the game away. AI, the professor insists, must move beyond curiosity and entertainment for the Ghanaian trader and provide useful information.

Every utility in history entered ordinary life through uses the serious people dismissed as trivial. Electricity entered the home as a light bulb — decades passed before it ran industry, refrigeration, and medicine. The internet entered as email and chat rooms, mocked as toys, before it carried commerce and government. The mobile phone entered as talk and texting before it became a bank. The trivial use is not a distraction from the transformative use. It is the on-ramp to it — the mechanism by which a population acquires the fluency, the trust, and the habits on which every serious application later depends.

The teenager in Tamale using AI for entertainment tonight is not wasting the technology. He is conducting the informal national training program that Wiafe's formal talent pipeline will one day harvest. Sneering at the light bulb has never once accelerated the arrival of the power grid.

Finally, the matter of the title — and here I address the professor not as a critic but as a practitioner who shares his destination and fears what his provocation will do once it leaves his hands.

“Why AI Is Irrelevant to Ghana” is a superb title for a lecture hall, where the audience stays for the argument and the word until does its careful work. It is a terrible title for a country, because a country does not stay for the argument. A country receives the headline. And the headline is already loose: a national newspaper of record has told several million Ghanaians that a professor of computer science has explained why AI is irrelevant to them.

Consider who now holds that headline. The ministry official looking for cover to defer a digitization budget. The school head who has been waiting for a respectable reason to ban the tools rather than teach them. The MP who needs one quotable authority to dismiss the AI strategy line item. None of them will watch the lecture. All of them can now cite it. In a policy environment where inaction is always the cheapest option, a credentialed “irrelevant” is not a provocation. It is an alibi — and it will still be circulating in committee rooms long after the nuance has been forgotten.

Anyone who holds a microphone in this debate — the professor at his lectern, the columnist at his desk — does not get to publish the provocation and disown the paraphrase. The headline is the policy input. Choose it the way you would choose a policy.

I am not waiting for AI to become relevant to Ghana. I am busy making it so.

Sources: Prof. Isaac Wiafe, “Why AI Is Irrelevant to Ghana: Reclaiming Our Future Through Human-Centred Transformation,” inaugural lecture, University of Ghana Great Hall, July 2, 2026; GhanaWeb reporting on same; GSMA, State of the Industry Report on Mobile Money 2026; Farmerline (farmerline.co); 3Farmate (3farmate.com); Xavier Space Solutions (xavierspacesolutions.com); minoHealth AI Labs (minohealth.ai).

Disclosure: the author's alumni association, the GSTS Alumni Association, has a partnership with Xavier Space Solutions on an educational satellite ground-station project.