Africa’s AI Declaration Is A Good Start, But Now Comes The Hard Part

On April 4, 2025, in Kigali, leaders from across the African continent gathered to sign the Africa Declaration on Artificial Intelligence, a sweeping vision document that reads like both a manifesto and a mission statement for the continent’s digital future. It’s bold. It’s comprehensive. It’s also very, very fluffy in parts.

Let’s start with what the Declaration gets right. Africa has every reason to take AI seriously. The promise of machine learning, automation, and predictive analytics is not just about joining the fourth industrial revolution. It’s about reshaping agriculture, improving health systems, streamlining government services, and, frankly, catching up. The Declaration’s commitments to developing local talent, data infrastructure, and AI governance mechanisms are the right areas to focus on. The idea of an African AI Scientific Panel and the proposed $60 billion Africa AI Fund are not just ambitious; they’re essential if Africa wants to play offense rather than defence in the global AI race.

But for a continent that thrives on pragmatism, the Declaration falls short on a key metric: grounded execution.

Take the talent development section. Yes, continent-wide AI curricula sound exciting. But who is going to develop these curricula? Who will fund the roll-out? What incentives exist to retain the talent once it’s trained? Africa’s brain drain problem won’t be solved by declarations. It requires concrete action on creating opportunity, not just education.

Or look at the part about compute infrastructure. The idea of “distributed sovereign compute infrastructure” connected by a “continental high-speed network” sounds great. But over half the continent still struggles with consistent electricity, let alone reliable internet. Ambition without realism becomes rhetoric.

Then there’s data. The declaration commits to “African open datasets” and “open AI models,” which is promising. But it glosses over the political, ethical, and technical complexities of data sharing in a fragmented, multilingual, and diverse region. Countries barely share health data during pandemics. Expecting seamless AI data collaboration across borders is a tall order without trust frameworks, legal harmonization, and deep capacity building.

Still, the Declaration is not without value. It signals to the world, and to Africa’s own policymakers, that AI is not just a Western or Asian game. It is a local issue. It affects farmers, students, doctors, and civil servants. The inclusion of terms like “human dignity,” “environmental sustainability,” and “intellectual property rights” shows that African leaders are thinking beyond just technology and toward values and outcomes. That’s commendable.

But the document also suffers from what too many continental strategies often do: a lack of prioritization. Everything is important, but not everything is urgent. Should the first order of business be setting up an African AI Council or getting GPUs into research labs? Should we rush to build a governance framework or start by funding five pan-African research centres of excellence? The Declaration doesn’t say.

And perhaps that’s the biggest weakness of all. This is a vision, not a plan. And Africa doesn’t need more visions. It needs execution.

What now? A few ideas:

1. Build regional AI labs, not committees. Choose three to five universities or institutions and turn them into AI powerhouses with real funding and partnerships. Start small. Build deep.
2. Create incentives for AI startups solving African problems. Use part of the proposed $60B fund to match investments in AI startups tackling local challenges, from language models for low-resource languages to crop disease prediction engines.
3. Open up public datasets, now. Start with anonymized health and agriculture data. No AI ecosystem can flourish without data, and governments are sitting on mountains of it.
4. Prioritize compute access before governance frameworks. Without compute, the rest is just talk. Negotiate bulk regional deals for cloud credits or set up shared HPC resources accessible to researchers and startups.
5. Forget the sandboxes for a second. Fix the basics. Laws, policies, and regulations are only as good as the institutions enforcing them. Invest in strengthening these institutions first.

The Africa Declaration on AI is a good first move. But history will judge us not by the declarations we sign, but by the actions we take after the cameras are gone. If Africa is serious about AI, it’s time to stop drafting documents and start writing code.