From Labour to Capital: The AI Shift South Africa Cannot Afford

A United Nations warning takes on unusual force in a country where more than three in every ten want work and cannot find it.

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://open.substack.com/pub/johanosteyn/p/from-labour-to-capital-the-ai-shift?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

At the beginning of July, the United Nations convened every member state in Geneva for the first Global Dialogue on Artificial Intelligence Governance, and its Independent Scientific Panel issued the strongest warning any international body has yet published on the technology. Buried inside that warning, alongside the language about catastrophic risk and children’s safety, is one sentence that ought to concentrate every South African board and cabinet mind. Without complementary investment in skills, workflows, infrastructure and labour-market institutions, the panel wrote, artificial intelligence risks widening inequality, displacing workers and shifting wealth from labour to capital. It is a general warning to every economy. In South Africa it lands with unusual force, because we are arriving at that shift with the fewest reserves in the room.

CONTEXT AND BACKGROUND

The Geneva meeting produced a rare consensus among governments, scientists and unions that artificial intelligence is outpacing scientific understanding and governments’ ability to adapt, that its distributional consequences will not correct themselves, and that the next session in New York in May 2027 must give organised labour a seat at the table it did not have in July. That warning arrived in a country whose first-quarter labour figures for 2026 recorded an official unemployment rate of 32.7 per cent, an expanded rate of 43.7 per cent, more than eight million people out of work, and a youth rate of 60.9 per cent for those aged 15 to 24 . We are, in other words, being asked to prepare for a labour-to-capital shift while more than three million young people already sit outside the workforce entirely.

INSIGHT AND ANALYSIS

The mechanism is well documented. Independent research shows that AI’s productivity gains flow disproportionately to workers already well-placed to use them, with the benefits highest for graduates and steadily declining as education levels drop, so a technology that raises average productivity can widen inequality even as it lifts the mean. At the market level, the same pattern repeats, with AI investment and returns concentrated in a handful of firms with the data, compute and distribution to capture them, so the value released by the technology drifts toward assets that are far more concentrated than human skills. In an already unequal economy, this is not a neutral trend.

I have previously written about this, arguing that in a country structured like ours the augmentation dividend flows to those already positioned to receive it, and a powerful general-purpose tool becomes a force for stratification unless deliberate policy intervenes. The United Nations panel now makes the corresponding policy claim explicit. Economic growth does not distribute itself; distribution is a governance choice.

IMPLICATIONS

South Africa is currently making that choice by default. The Draft National AI Policy was withdrawn in April after its reference list was found to contain sources fabricated by generative AI, and an expert panel chaired by Wits researcher Benjamin Rosman was appointed to rebuild it, so the country now faces the Geneva agenda without a framework of its own. That leaves the complementary investments the panel named, skills, workflows, infrastructure and labour-market institutions- without a coordinating instrument at the moment they are most needed.

For boards, the implication is direct. The productivity gains from adoption will fall to the firms that capture them, but the wider labour costs will fall to a society already stretched, and the political consequences of a shift that visibly moves wealth from workers to owners in this country will not stay quiet. Firms deploying AI at scale therefore have both a commercial and a governance interest in supporting worker transition, sectoral bargaining and skills investment, because a technology that arrives without a social settlement produces one afterwards, on worse terms.

CLOSING TAKEAWAY

The Geneva warning is not that artificial intelligence will inevitably impoverish workers. It is that whether it does depends on decisions being taken now about skills, institutions and bargaining, and that those decisions are being ducked. South Africa cannot afford to duck them. A shift from labour to capital in the world’s most unequal economy is the outcome least compatible with either social stability or long-run growth, and it is the outcome that arrives if government, business and organised labour treat AI as a productivity story rather than a distributional one. The panel gave the world a warning and a policy programme. Our task, before New York in 2027, is to answer it with a domestic settlement of our own.

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net