AI Governance for Africa > Part 2 > Section 3
As the previous section showed, AI governance is clearly a stated priority for Africa. But how far have domestic governments advanced in enacting AI governance?
One could argue that harnessing the power of AI aligns with Africa’s broader continental strategy. The AU Agenda 2063 envisages accelerated integration and technology across the 55 states.[1]
But the table below shows, there is still some way to go to meet this aspiration.
The salient trends on domestic governance include the following:
- No country has adopted full AI legislation (although at least two have adopted partial legislation, not including data protection laws).
- At least 31 African countries have adopted a data protection law that addresses the use of AI in relation to data: specifically, these 31 countries guard against automated decision-making.
- Relatively few have crafted national AI policies (8) or strategies (12).
- Only 15 African countries appear to have established an expert body on AI.
Read together, these findings suggest African countries have developed limited governance responses to AI. No country has adopted dedicated AI legislation, and the most prominent form of governance is enabled through data protection laws. Although very few countries have adopted a national policy, several of them have expert bodies or developed strategies.
Despite the limited measures, it is clear from continental instruments such as the AI for Africa Blueprint that AI is gaining traction as a policy concern. However, more is required to effectively govern AI in most countries.
How do data protection laws govern AI?
As the previous table shows, the most prominent form of AI governance in Africa is currently through data protection laws. This section briefly details how data protection laws regulate AI. Although there are some differences in domestic laws, data protection legislation generally does two things that concern AI: first, they include automated processing within the scope of their application and second, they provide a right against automated decision-making.
Automated processing: In the context of AI, automated processing involves the use of algorithms, rules, or instructions to perform tasks that would otherwise require human effort or decision-making. It can range from simple repetitive tasks to highly complex decision-making processes. It often involves the use of data and computational algorithms to analyse information, make predictions, optimize processes, and generate outputs. For example, automated processing can be used for data analysis where AI systems analyse large data sets to identify patterns or trends. It can be used for facial recognition where AI systems automatically process images to recognise objects or faces. These applications can have very real consequences – for example, they could determine a medical diagnosis or decide whether someone qualifies for a loan.
The right against automated decision-making: The right against automated decision-making is a fundamental aspect of data protection laws and is provided in most legislation around the world. The right is aimed at safeguarding individuals from potentially harmful or unfair decisions made solely by automated systems without human intervention. In the examples above, AI applications could perform analytic or recognition functions without any human involvement. This means that they could analyse all available information about a specific individual and determine whether or not they qualify for a loan. Such decision could have legal or consequential implications for an individual. Such decisions carry further risk as the results may be discriminatory or biased and would lack explainability.
The right provided in data protection law attempts to guard against these risks by providing individuals with a right not to be subject to such a decision. The right generally requires that a data subject be notified when such a decision has been made and allows them to request it to be reconsidered. The purpose of the right is to ensure that there is a degree of human oversight and involvement in these decisions.
References
Share
Navigate sections
Download Part 2
Explore the rest of the toolkit
Part 1: Introduction to AI Governance
Part 1 gives an overview of AI governance principles and approaches, and outlines international frameworks, with case studies from the European Union, the United States, and China. It discusses common concerns and themes driving AI governance.
Part 3: Advocacy Strategies for AI Governance
Part 3 explores a series of key questions for the design of advocacy strategies on AI governance, particularly in African contexts.
