Nearly 8 months after its introduction, South Africa’s National Artificial Intelligence Policy (AI) has been a success. The Department of Communications and Digital Technologies will be reviewing the policy framework in April. Experts are concerned about the slow progress of the framework, which promotes ethical and inclusive AI development, talent development, innovation and data protection.
Nerushka Bowan is a technology and privacy attorney and founder of LITT Institute. She said, “At this time, there’s a lot uncertainty.” “The biggest challenge would be to provide a clearly defined and actionable roadmap that guides the country’s AI policies and regulatory intent.” It is still lagging behind in the implementation of its AI policy due to an unclear direction. Daniel Novitzkas is the group director of Specno – a South African digital solution company. He said that without clear regulatory guidance the country risks falling further behind peers such as Rwanda and missing out on opportunities for economic transformation and international competitiveness. Rwanda’s national AI policy was approved in April 2023. It focuses on using AI to improve public services and drive economic development. It also prioritizes ethical AI development, incentives for investment, and infrastructure expansion. South Africa could miss out on the estimated ahref=””https://repository.uneca.org/handle/10855/50180″ “rel=””noreferrer noopener””target=””_blank” “>$1.5 billion contribution that AI will make to Africa’s GDP in 2030. Wendy Rosenberg, director and head of digital media, electronic communications and governance at Werksmans Attorneys said that South Africa’s AI Framework covers critical areas such as data protection, privacy and transparency. However, the delay is problematic. Rosenberg said that these issues are crucial in ensuring AI deployment and development align with South Africa’s ethical and legal landscape. It is important to finalise the framework policy as it will set the foundation for the detailed policies that are established for different sectors. Bowan said that there is currently no AI-specific regulation, or guiding policy, to provide clarity about government intentions and future legal obligations.
Without an AI framework policy that is supported by the government, it becomes more difficult to encourage entrepreneurs and to attract foreign investment into AI infrastructure. Investors and developers are often looking for regulatory clarity before investing resources. The absence of a definitive AI policy creates hesitancy, which stalls AI-driven economic change.
Bowan said, “We don’t have many years left to debate the future.” Investors want to see the landscape. Early movers gain an advantage.
Lack of clarity not only deters investors, but it also risks driving local AI talents abroadwhere countries with well defined AI policies such as the U.S.A., U.K. and Canada actively attract skilled professionals by offering funding incentives, grants for research, and AI friendly regulations.
The people who can build AI solutions for our economic growth need the right information, tools, and regulations in order to thrive. Novitzkas said that if we don’t, we could lose them to the US and Europe.
Addressing concerns about bias and ethical issues
South Africa has a solid foundation for AI regulation thanks to the Protection of Personal Information Act, which is aligned with the General Data Protection Regulation of the European Union (GDPR). Rosenberg emphasized that AI policies should go beyond data security to address transparency, bias reduction, and ethical AI usage. Rosenberg stated that AI systems use personal data for various purposes, including training AI models, personalisation and analytics. Transparency and user control are therefore vital.
Addressing ethics concerns in AI-bias and transparency is crucial, especially in South Africa, with its history of inequalities. Rosenberg stated that global best practices, such as human in the loop systems, bias evaluation process, and diverse data samples, should be implemented to reduce these risks.
Risk-based approach
In order for South Africa to fully capitalize on AI, the country must address key issues, including data sovereignty and internet accessibility. About 28% of South Africans will not have internet access by 2023. This will limit the country’s capacity to fully harness AI. Novitzkas stated that “even if AI can address pressing issues such as education, the average South African does not have access to the internet or smartphone.”
The European Union’s AI Act, which is based on a risk-based system, has stricter regulations for high-risk applications, while allowing greater flexibility for low-risk AI. A phased approach, along with ongoing legislative updates, will be required for AI policy implementation.
The challenge is always the law keeping pace with technology. Rosenberg said that we need a future-ready, principle-based legislation.
Regulations are necessary, but they must not stifle the innovation. If South Africa can find the right balance between oversight, technological growth and investment, AI could be a major driver for investment, job creation and digital transformation.
