AI in Africa: 5 issues that must be tackled for digital equality

First Published,September 25,The Conversation

Author-Rachel Adams

Honorary Research Fellow of The Ethics Lab, University of Cape Town

If it’s steered correctly, artificial intelligence (AI) has the potential to accelerate development. It can drive breakthroughs in agriculture. It can expand access to healthcare and education. It can boost financial inclusion and strengthen democratic participation.

But without deliberate action, the AI “revolution” risks deepening inequality more than it will expand opportunity.

As a scholar of the history and future of AI, I’ve written about the dangers of AI widening global inequality. There’s an urgent need to develop governance mechanisms that will try to redistribute the benefits of this technology.

The scale of the AI gap is stark. Africa holds less than 1% of global data centre capacity. Data centres are the engines that drive AI. This means the continent has minimal infrastructure for hosting the computing power necessary to build and run AI models.

While only 32 countries worldwide host specialised AI data centres, the US and China account for over 90% of them.

And only about 5% of Africa’s AI talent (innovators with AI skills) have sufficient access to the resources needed for advanced research and innovation.

But they also still have the chance to set a new trajectory – one where Africa and the global majority shape the rules of the game. One that ensures AI becomes a force for shared prosperity rather than exclusion.

To achieve this, five critical policy areas most be addressed. These are data; computing capacity; AI for local languages; skills and AI literacy; and AI safety, ethics and governance. These are not just African priorities; they’re global imperatives.

1. Compute and infrastructure

Access to computational power has become the defining chokepoint in today’s AI ecosystem. African researchers and innovators will remain on the margins of the AI economy unless there is investment in regional data centres, GPU clusters (a group of computers working together on large-scale AI processing) and secure cloud infrastructure.

Europe, by contrast, has pooled over US$8 billion in establishing the European High-Performance Computing Joint Undertaking to ensure the continent has computing capacity for local innovations.

African countries should press for funding and partnerships to expand local capacity. They will also need to insist on transparency from global providers about who controls access, and ensure regional cooperation to pool resources across borders.

2. Data governance

AI systems are only as good as the data they’re trained on. Much of the continent’s data is fragmented, poorly governed, or extracted without fairly compensating those it’s collected from. Large, diverse and machine-readable datasets are used to teach AI models about the contexts and realities the data reflect.

Where ethical stewardship frameworks exist, locally managed datasets have already driven innovation that has impact. For example, the Lacuna Fund has helped researchers across Africa build over 75 open-machine-learning datasets in areas like agriculture, health, climate and low-resource languages. These have filled critical data gaps, allowing for tools that better reflect African realities. Realities like high-accuracy crop yield datasets for farming. Or voice/text resources for under-served languages.

Robust national data protection and governance laws are needed. So are regional data commons, a shared resource where data is collected, stored, and made accessible to a community under common standards and governance. This would enable collaboration, reuse, and equitable benefits. Standards for quality, openness, interoperability and ethics developed by multilateral organisations must be designed with African priorities at their centre.

3. AI for local languages

Inclusive AI depends on the languages it speaks. Current large models overwhelmingly privilege English and other dominant languages. African languages are all but invisible in the digital sphere. This not only entrenches existing biases and inequalities, it also risks excluding millions from access to AI-enabled services.

Take the example of the Cape Town-based non-profit organisation Gender Rights in Tech. It has developed a trauma-informed chatbot called Zuzi that supports survivors of gender-based violence by providing anonymous, accessible guidance in diverse South African languages on their rights, available legal services, and sexual and reproductive health. It helps overcome stigma and bridge gaps in access. Zuzi demonstrates the power of AI technologies in local languages.

Dedicated investment in datasets, benchmarks, and models for African languages is urgently needed, as well as in tools for speech recognition, text-to-speech, and literacy.

4. AI skills and literacy

African infrastructure and data will mean little without human capacity to use them. At present, AI skills supply falls far short of demand, and public understanding of AI’s benefits and risks remains low.

To increase skills, AI and data science will need to be integrated into school and university curricula, and vocational training will need to be expanded. Supporting lifelong learning programmes is essential.

Public awareness campaigns can ensure citizens understand both the promise and perils of AI. This will support deeper public debate on these issues. It can also target support for women, rural communities, and African language speakers to help prevent new divides from forming.

5. Safety, ethics, and governance

Finally, stronger governance frameworks are urgently needed. African countries face unique risks from AI. Among them are electoral interference, disinformation, job disruption, and environmental costs. These risks are shaped by Africa’s structural realities: fragile information ecosystems, large informal labour markets, weak social safety nets, and resource-strained infrastructure. National strategies are emerging, but enforcement capacity and oversight remain limited.

African governments should push for the creation of an African AI safety institute. Safety and ethical audits must be mandated for high-risk systems. Regulations and AI governance instruments must be aligned with rights-based African principles that emphasise equity, justice, transparency, and accountability. Participation in global standard-setting bodies is also crucial to ensure that African perspectives help shape the rules being written elsewhere.

All eyes on the G20

Taken together, these priorities are not defensive measures but a blueprint for empowerment. If pursued, they would reduce the risk of inequality. They would position Africa and other regions across the majority world to shape AI in ways that serve their people and economies.

Digital and technology ministers from the world’s biggest economies will be attending the G20’s digital economy working group ministerial meeting at the end of September.

On paper, it’s a routine meeting. In practice, it may be the most consequential gathering on AI policy Africa has ever hosted

This is the first time the G20’s digital ministers are meeting on African soil. It’s happening at the very moment AI is being hailed as the technology that will redefine the global economy.

This meeting will not stand alone. It will be followed by the AI for Africa conference, co-hosted by South Africa’s G20 presidency, Unesco and the African Union. Here the AI in Africa Initiative will be launched. It is designed as a practical mechanism to carry forward the G20’s commitments and advance implementation of the African Union’s Continental AI Strategy.

Cape Town could mark a turning point: the moment when African leadership, working in concert with the G20, starts to close the AI divide and harness this technology for shared prosperity.

SOURCE

The Conversation