SA’s AI governance dilemma

Prof Johan Steyn, AI thought leader / Avashlin Moodley, AI strategist.
Prof Johan Steyn, AI thought leader / Avashlin Moodley, AI strategist.

South Africa is currently navigating a complex landscape in the development of an AI regulatory framework, a challenge highlighted by AI specialists Prof Johan Steyn and Avashlin Moodley during a recent webinar hosted by CTU Training Solutions. As prominent figures in the AI community, Steyn and Moodley emphasised the need for a uniquely African approach to AI governance, stressing that the continent's diverse needs and contexts must shape its regulatory frameworks.

Current state of AI governance in South Africa

According to Prof Steyn, South Africa lags in establishing a comprehensive AI regulatory framework. He points out that, much like the EU's GDPR, which has set a high standard for data privacy but required adaptation for South African application in the form of POPIA, the EU AI Act will guide South Africa in developing its own set of measures.

Moodley points out that the EU AI Act can’t be enforced within the boundaries of SA, but if you try to do business internationally, you’ll need to abide by international laws.

Steyn underscores the delicate balance required between regulation and innovation, cautioning that over-regulation could stifle technological advancements.

One notable aspect of the proposed EU AI Act is its emphasis on explainable AI, particularly in automated decision-making systems. In South Africa, such systems are already in use in sectors like banking, where automated credit decision-making is common. The EU's approach mandates that consumers be informed if an algorithm influenced a decision, providing transparency and accountability. Implementing similar measures in South Africa could pose challenges, including the potential burden on customer service infrastructure such as call centres, says Moodley.

African perspective and regional collaboration

Both Steyn and Moodley advocate for a regulatory framework tailored to the African context. They argue against a one-size-fits-all approach, suggesting that individual African countries should have the flexibility to adapt regulations to their unique circumstances. This perspective is critical given Africa's diversity and the historical imposition of global standards that do not consider the continent's specificities.

The speakers highlight the importance of African solutions for African problems. With countries like Kenya, Rwanda and South Africa at the forefront of AI development, there is a unique opportunity for regional collaboration. “A collective effort could foster a robust, adaptable framework that reflects the diverse cultural and economic landscapes across the continent,” says Moodley.

Challenges and the way forward

The path to developing an AI regulatory framework in South Africa is fraught with challenges. As mentioned previously, one of the significant concerns is the speed at which AI technologies evolve, often outpacing the legislative processes needed to regulate them. The recent release of a draft national AI plan in South Africa signals a step forward, but implementation remains slow. The formation of a working group, comprising representatives from the National Advisory Council on Innovation, the Council for Scientific and Industrial Research, the Human Sciences Research Council and the Department of Science and Innovation aims to address these issues. This group is tasked with creating a recommendation framework for the President and the Cabinet, with the goal of producing actionable and impactful policies.

Prof Steyn stresses the importance of involving all stakeholders in this process, including educational institutions, tech companies, startups and government bodies. He believes collaboration across these sectors is crucial for developing a comprehensive and effective regulatory framework. “Hopefully with a larger collaborative effort, we’ll be sure to not just copy and paste what the EU is doing.”

Conclusion

As South Africa and the broader African continent move towards establishing AI governance frameworks, it is clear that a nuanced, context-specific approach is necessary. Learning from global standards, like the EU AI Act, while ensuring these frameworks are adaptable to African realities, will be key. The involvement of a diverse group of stakeholders will help shape a balanced and inclusive regulatory environment that promotes innovation while protecting citizens' rights and interests.

For more insights from Prof Steyn and Moodley, you can watch the full AI governance webinar on CTU Training Solutions’ YouTube channel here.

Share