The GNU must prioritise artificial intelligence at all levels of government

The government of national unity (GNU) is formulating its strategic priorities and action plans now that the electoral dust has settled. The GNU has pressing challenges, such as those prioritised by the much-lauded Operation Vulindlela. Addressing these structural obstacles to growth are essential to South Africa’s development.

However, the GNU should not only focus on legacy issues, but also address novel and emerging opportunities and challenges. Arguably, chief among these is the secular rise of artificial intelligence (AI).

AI’s progression marks a turning point, being a general-purpose technology (GPT) which permanently transforms economic systems. Other examples of GPTs include steam engines, electricity, computers and the internet.

Given its GPT status, AI’s reach will continue to expand at an exponential rate, influencing more areas and to greater extents. It will affect almost all areas of the government's focus, including education, health, employment, national security, intelligence, international relations, crime and social welfare. Not to mention its impact on macroeconomic stability and financial systems.

In this context, the GNU should integrate AI into its apex priorities, recently outlined by President Cyril Ramaphosa. The GNU’s three strategic priorities are: first, to drive inclusive growth and job creation; second, to reduce poverty and tackle the high cost of living; and to build a capable, ethical and developmental state.

AI can help achieve these goals and its growing ubiquity can put upward or downward pressure on these priorities. For instance, IMF research claims that AI holds the potential to boost productivity growth and advance public service delivery, but it also raises profound concerns about massive labour disruptions and rising inequality.

Given AI’s wide and growing purview, the South African government cannot afford to keep the technology on the fringes of policy discussions, especially as there are AI country-specific risks.

There is a range of ways that the government can incorporate AI into its operations and decision-making. For now, however, AI hardly features on the country’s political agenda. For instance, it was barely mentioned during the recent election season.

More broadly, there are few examples of policymakers giving little more than lip service to AI – most remarks have been reductionist platitudes. In contrast to this policy-platitudes vacuum, the South African population and business sector’s interest and use of AI continues to grow.

This borderline AI policy inertia must be reversed. What is needed instead is a bespoke, strategic and contextually focused plan for how the country can use AI and, in turn, how AI may affect the country.

In short, the GNU needs to redouble its efforts to define an AI strategy. Efforts so far have been stuttered and underdeveloped. Earlier in 2024, the Department of Communications and Digital Technology (DCDT), the government’s de facto custodian of AI, released a discussion document that sets the path for an AI policy.

The DCDT’s effort should be welcomed as a much-needed genesis of an eventual AI strategy, but there is significant room for improvement. On the positive side, it is heartening that the government is taking concrete steps towards an AI policy – better late than never – but the country is still behind other middle-income countries.

There is also an attempt to integrate AI within the existing policy framework, such as the National Development Plan. On the negative side, however, the document is repetitive, vague and contains conflicting provisions. Moreover, large parts of the document are generic, but simultaneously contain specific and apparently arbitrary targets and goals. In its current state, it is difficult to envision how this document will be developed into a holistic strategy and action plan.

In fairness, the DCDT did note that the policy will be further progressed by an AI Expert Advisory Council. Additionally, the Artificial Intelligence Institute of South Africa (AIISA) – self-described as the “brainchild” of the DCDT, the University of Johannesburg and the Tshwane University of Technology – will execute the plan.

Unfortunately, neither the DCDT nor the AIISA have provided any update on either the council or the policy’s development. In fact, the AIISA’s website has not even been updated since March 2023. None of this augurs well for an eventual AI strategy, which is supposed to be inclusive and have broad stakeholder buy-in.

The DCDT’s new minister, Solly Malatsi, who heralds from the DA, should make the development of the AI strategy one of his apex priorities. Moreover, the country’s AI plan cannot simply be the purview of the DCDT alone – a whole-of-government approach is needed.

The AI strategy needs to be cross-functional, holistic and account for the full spectrum of government in terms of levels (ie, national, provincial, local) and functional areas. Artificial intelligence should be approached as a cross-cutting issue – similar to climate change – and not relegated to a single department or addressed by government entities in an ad hoc manner.

South Africa can ill afford five more years of glacial progress on AI. With fresh talent and diverse political perspectives, the GNU offers the nation a renewed opportunity to harness AI as a catalytic force, positioning the country to capitalise on AI opportunities while mitigating its risks. DM