Government Set to Revive National AI Policy
The government has announced its plans to relaunch the national artificial intelligence (AI) policy. The previous iteration merely recited aspirations and responsibilities without providing a comprehensive strategy. The challenge now lies in whether the updated policy will truly define the governance structure, establish legal bases, and outline the consequences for non-compliance.
Clarifying Regulatory Authority
The proposed policy assigns roles to the South African Independent Communications Authority, the Information Regulator, and the Competition Commission, which is a positive development. However, designating a role does not necessarily equate to active governance. These institutions ought to be obligated to publicly clarify how their existing legal frameworks apply to AI systems in their jurisdictions, as well as their intended actions for the coming year. Implementing this requirement will be cost-effective and lead to quicker, more impactful governance than merely crafting new strategic documents or creating additional organizational structures.
Focusing on Specific Technologies
Rather than trying to encompass the broad term “AI,” regulation should focus on individual technologies. The European Union took four years to draft an extensive AI law that categorizes risks into four levels, but its enforceability remains uncertain. This model, which inspired South Africa’s draft policy, may not be suitable given the country’s differing regulatory capabilities. Conversely, China’s approach—regulating technologies sequentially, like recommendation algorithms and synthetic media—offers a more pragmatic route. South Africa should prioritize synthetic media regulation, especially given the potential disruptions deepfakes could cause in a country facing political instability. Following that, regulations should address algorithmic decision-making in critical areas such as credit and employment.
Learning from Economic Historians
Insights from economic historians, rather than futurists, can offer a more grounded perspective on AI’s impact. Daron Acemoglu, a Nobel Prize-winning economist, explores the disparity between wealthy and impoverished nations in his recent work, “Power and Progress.” He argues that benefits derived from technology often accrue to those who control it unless political systems are designed to redistribute these advantages. Historical instances, from mechanized textiles to electrification, highlight this dynamic. AI governance must confront this challenge, and South Africa can leverage competition policy to create mechanisms for equitable benefit distribution.
Collaborating within the AfCFTA Framework
The draft policy recognizes the importance of aligning with the African Continental Free Trade Area (AfCFTA) and broader continental goals, but these references remain largely aspirational. It is crucial to harmonize the domestic AI regulatory framework with commitments to AfCFTA, as developing AI responsibly in Africa requires a market significant enough to attract investment. As Africa’s most advanced economy, South Africa has both a responsibility and a vested interest in spearheading initiatives for collaborative data governance and accountability standards across the continent.
Utilizing National Procurement for Regulation
One critical mechanism absent from the previous draft—yet vital for immediate impact—is the integration of national digital procurement processes with regulatory requirements. Government contracts should be restricted to companies affiliated with accredited industry associations, mandating them to adhere to jointly established technical and ethical standards. This approach shifts compliance costs from the state to the industry. Firms seeking public contracts would need to engage with these industry bodies, and non-compliance would exclude them from contract opportunities. Such a strategy minimizes the perceived burden of regulations while empowering the industry to define its standards.
Rethinking Policy Reboot Strategies
In revamping policy, the temptation often exists to limit disruption with each iteration. However, the recent issues surrounding AI misuse highlight the need for a fundamental reevaluation of how these policies are conceived. This is an opportune moment to address these concerns comprehensively, ensuring the newly revised policy is more than just a recycled document.
Timcke is a senior fellow at the Center for Social Change at the University of Johannesburg and is affiliated with the Center for Information Technology and Public Life at the University of North Carolina at Chapel Hill.
