Member Commentary Series

Africa needs AI training. It needs more of it, not less.

Leaders need to understand what they are approving. Public servants need practical skills. Regulators need to understand the systems they are expected to govern. Universities need stronger learning pathways. Enterprises need people who can use AI responsibly. And across the continent, trainers and faculty need to be developed so that knowledge does not always have to come from outside.

Training matters. But training should not end with exposure. For training to serve sovereignty, capability must grow from it.

Across Africa, we have seen a familiar pattern. A workshop is delivered. Certificates are issued. Reports are written. Participants leave with new vocabulary, new awareness, and sometimes real enthusiasm. But months later, the institution itself may not be much stronger.

I have seen this across years of international development work. Capacity-building activities can be useful, well-intentioned, and professionally delivered, but still fail to leave enough capability behind.

That is the deeper challenge for Africa's AI sovereignty agenda. The question is not only whether Africans are being trained in AI. The question is whether African institutions are developing the internal capability to understand, govern, procure, adapt, deploy, improve, and ethically manage AI systems on African terms.

That is the heart of Talent Sovereignty. Africa cannot own its AI future without owning its AI workforce. But owning that workforce means more than producing trained individuals. It means building, retaining, deploying, and continuously strengthening African capability where real decisions are made.

The sovereignty questions inside adoption

AI adoption is already moving faster than institutional capability. Inside many institutions, some practical questions remain unresolved: What exactly are we adopting? Whose data is being used? Who understands the model? Who carries the risk? Who negotiates with vendors? Who audits the system? Who protects citizens? Who captures the value? Who remains dependent?

These are not only technical questions. They are sovereignty questions. An institution may look modern because it has adopted AI tools. But if it becomes more dependent on external consultants, external platforms, external infrastructure, external standards, and external interpretation of its own problems, then something important has not changed. That is not sovereignty. Sovereignty requires the ability to make informed choices.

What training must leave behind

Good AI training should leave people able to act differently inside their institutions. They should be able to question a vendor, read risk, connect technology to procurement and accountability, understand data governance, and relate AI to public value. Some should be able to teach others. Some should move from awareness into responsibility.

That difference matters. A regulator may attend training and still struggle to turn principles into oversight. A public servant may attend a workshop and still not know how to apply what was learned in service delivery. A leader may hear about AI and still outsource every strategic decision. A professional may learn new tools and still have no pathway into a role where that knowledge can serve African institutions. And if trainers are always imported, while local trainer capacity remains thin, then capability is still not accumulating where it should.

Talent Sovereignty asks us to raise the standard of what training must achieve. Training should build African trainer capacity. It should strengthen institutional readiness. It should connect professionals to real deployment pathways. It should help institutions govern, procure, adapt, and improve AI rather than merely consume it.

Partnership without dependency

This is not an argument against external partnership. Africa will continue to learn, collaborate, exchange knowledge, and work with global partners. That is necessary and valuable. But collaboration is not the same as dependency.

The test of a good AI capacity-building programme is what it leaves behind. Are African trainers stronger after it? Are institutions more ready? Is procurement judgment better? Is governance capability deeper? Are there clearer pathways for deployment? Are African institutions better placed to capture value? If the answer is no, then training may have happened, but Talent Sovereignty has not advanced far enough.

From training events to capability systems

The work ahead is not to choose between training and capability. Africa needs training precisely because it needs capability. The shift is from training as an event to training as a capability-building system — from exposure to competence, from workshops to institutional readiness, from imported expertise to African trainer capacity, from certificates to deployment, from adoption to value capture, from training delivered to capability retained.

Africa's AI sovereignty will not be secured by technology alone. It will depend on whether African institutions can build, retain, deploy, and continuously strengthen the human capability required to shape technology on African terms.

That is the work of Talent Sovereignty — and the contribution I hope to make through Pillar 4: helping shift the conversation from training delivered to capability retained, deployed, and owned.

Artificial intelligence is reshaping economies, institutions, and societies across the world. For Africa, the central issue is not simply whether the continent adopts AI, but whether African nations, institutions, and leaders have the capacity to shape the terms of that adoption.

AI development is moving quickly. Governments are drafting strategies. Companies are deploying new tools. Universities are training researchers and practitioners. Civil society organisations are raising important questions about accountability, inclusion, safety, rights, and public trust. Across the continent, important work is already underway.

But much of that work remains fragmented. Africa needs stronger networks that bring together policymakers, technologists, researchers, entrepreneurs, lawyers, educators, funders, development institutions, and civil society leaders. No single institution can manage the AI transition alone. The opportunities and risks of AI cut across sectors, borders, languages, economies, and disciplines.

An AI sovereignty network can help connect leaders who are working on related problems but may not yet be in sustained conversation with one another. It can support shared learning, amplify African expertise, identify common priorities, and help translate technical developments into policy, governance, investment, and public-interest outcomes.

AI sovereignty also requires more than aspiration. It requires institutions capable of building trust, coordinating action, developing standards, supporting talent, strengthening public-sector capacity, and helping leaders make informed decisions under conditions of rapid technological change. A network can serve as connective tissue across these efforts.

It can create space for ministers, regulators, executives, researchers, investors, academics, and civil society leaders to engage one another around shared priorities. It can help ensure that national strategies are informed by regional knowledge and that regional conversations remain connected to the practical needs of countries, organisations, and communities.

An African AI sovereignty network can also help ensure that global conversations about artificial intelligence include African perspectives from the beginning. Too often, rules, standards, and norms are shaped elsewhere and later applied to African markets and institutions. Africa should not only be a consumer of AI systems or a recipient of outside policy models. It should help shape the principles, practices, and partnerships that guide AI's development and use.

Building that influence requires coordination. It requires platforms where leaders can exchange ideas, develop talent, support responsible adoption, and build durable relationships across public, private, academic, financial, and civic sectors. It requires institutions that can convene, translate, and connect.

As AI adoption accelerates, Africa has an opportunity to build networks that match the scale of the transition. Strong networks can help turn scattered activity into shared capacity, and shared capacity into continental influence.

AI is transforming the world. Africa's ability to shape that transformation will depend not only on innovation, but also on collaboration, trust, institutional strength, and the networks built today.

There is an East African proverb that captures the infrastructure sovereignty challenge precisely: one finger cannot wash a face. You need the whole hand. Africa's AI infrastructure challenge is the same. The compute, data centre capacity, and high-speed connectivity required to build and run sovereign AI systems at scale cannot be delivered nation by nation. It requires collective architecture.

Today, the vast majority of AI computation affecting African institutions, governments, and citizens runs on infrastructure owned by a small number of foreign technology corporations. African data flows to foreign servers. African decisions are processed by foreign models. African intelligence is generated on foreign infrastructure — and the value it creates largely flows outward.

This is not a technology problem. It is a sovereignty problem. Infrastructure sovereignty — Pillar 2 of AISN's framework — is the recognition that owning your AI future requires owning, or having sovereign access to, the physical and digital infrastructure on which that future runs.

The path forward requires three parallel investments. First, shared continental data infrastructure — pan-African data centres that are owned, governed, and operated under African institutional frameworks. Several are already being built, including by AISN Network Members. But isolated efforts are insufficient. Second, collective compute access — negotiated access to high-performance compute capacity that African research institutions, governments, and enterprises can use to train and run AI models without depending on foreign cloud providers at foreign prices. Third, connectivity sovereignty — ensuring that the networks through which African data travels are not themselves chokepoints of external dependency.

None of these can be achieved by individual nations acting alone. They require the kind of institutional coordination, shared standards, and collective investment architecture that only a continental institution can provide. That is the infrastructure sovereignty mandate of the AI Sovereignty Network.

The Nigeria Data Protection Act (NDPA) 2023 represents a significant step forward in Nigeria's regulatory architecture. It establishes a legal basis for data protection, creates the Nigeria Data Protection Commission as an independent enforcement body, and sets out rights for data subjects that did not previously exist in statute. For a nation of over 200 million people and the largest technology ecosystem in Sub-Saharan Africa, this matters.

But the NDPA was designed primarily for the data economy that exists today — not the AI-driven economy that is emerging. It addresses data collection, processing, and transfer. It does not address algorithmic decision-making, automated profiling, AI-specific risk assessment, or the accountability gap that opens when AI systems make consequential decisions about Nigerian citizens without human oversight.

Regulatory sovereignty requires more than a data protection law. It requires the institutional capacity to govern AI systems deployed in Nigerian markets — including systems built, trained, and governed outside Nigeria — and to hold their operators accountable under Nigerian law. The NDPA is a foundation. It is not yet a framework.

Three gaps are particularly urgent. First, algorithmic accountability — the NDPA does not establish a right to explanation or contestation of automated decisions that affect Nigerian citizens. Second, AI-specific liability — when an AI system causes harm in Nigeria, the current legal framework does not clearly assign liability to the developer, deployer, or operator. Third, enforcement capacity — the Nigeria Data Protection Commission is nascent, and its ability to investigate complex AI governance failures at scale is not yet established.

These gaps are not unique to Nigeria. They reflect a continental challenge: Africa has enacted data protection laws faster than it has built the institutional capacity to enforce them, and AI governance demands capabilities that exceed what those laws were designed to deliver. The work ahead is not just legislative. It is institutional.