How can we best use AI in African public policy?

The use of AI presents enormous opportunities for governance. But to reap the benefits, work needs to be done now to ensure Africa is ready to harness this new technology, writes Gedion Onyango.

The use of Artificial Intelligence (AI) is now an undeniable reality for Africa, and its applications are set to expand across sectors in the coming decade. This will have important implications for public policy. Rwanda, Ghana, Algeria, Mauritius, Benin, Egypt, South Africa, Tunisia, Senegal, and Nigeria already have national policy frameworks or strategies for AI in government. Others, such as Botswana, Namibia, and Kenya, have begun developing AI strategies to guide its adoption in governance. It is expected that more countries will join this trend in response to the African Union (AU) and other continental organisations’ efforts to leverage AI for public service.

AI as a tool

In July 2024, the AU adopted its AU Strategy on Artificial Intelligence. This mirrors the European Union’s AI Act 2024 and is likely to influence discussions on national AI policies across member states. The main goal is to promote a unified approach to AI adoption and governance that fosters data-driven innovation and value creation, while addressing risks like data inequality, weak infrastructure, and poor data governance.

The AU’s policy encourages sector-specific approaches and establishes the essential conditions for safe AI deployment. These conditions include defining data types, classifications, and infrastructure needs to ensure trustworthy data systems tailored to local contexts. It advocates for collaboration between African governments and NGOs to incorporate AI into various sectors—including immigration, healthcare, climate change, and education—to improve public services and policymaking.

In other sectors, AI could be used to steer disruptive innovation and profitability. But for governments, AI should be leveraged in a manner that does not erode values that embody what public administration is, despite the potential of AI to revolutionise  how public administration functions. AI for government should play a cumulative function without threatening democratic political gains and citizen agency mechanisms. These gains and mechanisms embody the core values of democracy, including representation, equality, equity, accountability, responsibility, responsiveness, fairness, and justice. AI for government should promote these values, in their legal, political, and managerial aspects in public administration. AI should play a complementary role as a tool, not as an actor or an alternative administrative structure.

The key question is whether African countries are prepared for AI and capable of utilising it to maximise its policy benefits without worsening issues like human rights violations and inequality. 

To this end, effective AI governance will rely on solid data infrastructure that is anchored in appropriate legislation and principles of privacy. These are the fundamental principles that should shape the use of AI for government.

What AI frameworks will work for Africa?

The creation of several AI Labs across major African cities by corporations like Google and IBM Labs in Johannesburg, Accra, and Nairobi shows that African countries are actively exploring the potential benefits of AI. However, many African governments, like their global counterparts, have yet to establish clear guidelines for AI deployment. Additionally, the digital infrastructure in most African countries is still developing and cannot yet support effective AI systems. The risks tied to AI use could also threaten Africa’s fragile regulatory state and may worsen existing problems of inequality and human rights violations. Common policy issues such as the lack of context-specific global policies, poor digital infrastructure, limited national investment in AI, and algorithmic bias remain significant concerns.

AI frameworks should address what AI for African society and governance should look like. The focus must be on exploring ethical issues specific to AI’s implications for Africa. For instance, how does AI’s algorithmic bias impact racial relations, gender, and geographical inequalities? To address this, government AI systems should be grounded on local values, reducing anti-African and anti-Black biases, and be tailored to the needs of specific policy targets.

Sound AI governance mainly depends on effective data governance and governance practices. Therefore, while adopting AI templates from successful models like the EU is convenient, it exacerbates global policy convergence, which often hinders African governments from customising governance to their own contexts. Widespread convergences are likely to prevent the widespread adoption of AI from meeting Africans’ needs.

To make the most of the opportunities, African AI strategy frameworks should adopt a proactive rather than reactive approach to public policy. This should steer the creation of sustainable and culturally relevant AI government systems. Starting today, African governments should remodel their systems to embed AI laws and regulations. This should promote responsible innovation and digitalisation processes that support ethical and secure AI. This should emphasise clear explanations of algorithmic decisions, data protection (privacy-by-design), and hold developers accountable for ethical data use (Transparency and accountability). Furthermore, ensuring interoperability within AI architecture for government operations is crucial.

A holistic, collaborative assessment is needed to understand the complex interplay of moral, ethical, technological, and legal principles, as well as political and managerial values related to the use of AI by governments. This should be guided by the Civic Technology community to help shape future AI narratives.

How to achieve AI in African public policy

There are many opportunities for the effective use of AI for government, but these are laced with critical gaps and challenges. An effective implementation model will need to incorporate capacity building and skills training for policymakers and all other actors involved. Such projects should aim to address deep-seated barriers due to institutional culture and historical inequalities in the public sector.

Leveraging AI for government should go beyond digital infrastructural investment to address the inconsistencies of the state’s approach to digital transformation. The implementation design should create co-jurisdictional, cooperative, and collaborative frameworks. This may need sector-to-sector evaluation mechanisms of AI readiness of public organisations and other sectors involved in public service delivery.

A robust monitoring and evaluation mechanism anchored in learning and co-development processes will ensure co-ownership and sustainable adaptation of AI strategies in government institutions. Similarly, regulatory mechanisms are essential components of implementation to regulate private entities using AI, ensuring effective accountability procedures. Sound regulatory mechanisms will guide stakeholder engagements and foster cooperation aimed at reducing jurisdictional conflicts and priorities common in public sector collaborations.

To leverage AI for government, due care and attention must be paid to fit-to-context frameworks and regulations rather than off the shelf solutions. Leveraging AI requires caution so as not to undo the political gains that guide governance and consideration for human rights values going forward. The AI revolution is here. We need to focus on how to best use it to improve public administration, and we need to do so now.

This blog is based on the book Designing Artificial Intelligence for Public Policy and Governance in Africa edited by Gedion Onyango.