Quick Summary

AI is the most powerful cost-cutting instrument in history. But if every company uses it to reduce headcount, who is left to buy what they produce? This piece makes the economic case for social innovation alongside technological innovation and what South African employers must consider before implementing.

Reading time: 6 minutes.

There has never been a better time to cut costs. That is precisely the problem.

AI is now offering companies something they have always wanted dramatically lower costs, at scale, with minimal friction. The productivity gains are real. The efficiency argument is sound. The business case for automation writes itself.

But before you reach for the saw remember this.

A man sits on a branch, sawing the wood between himself and the trunk. He is working hard. Making progress. He cannot see what is about to happen. Next to him is a mirror. It is not showing the future. It is showing the consequence of what he is already doing.

The growth illusion

For several decades, real wage growth for ordinary workers in most economies has lagged well behind productivity growth. Economists including Thomas Piketty and researchers at the OECD have documented the gap: the top 10% of earners have seen significant real income gains while the bottom 50% have experienced stagnation. The appearance of broad-based growth has been sustained through credit expansion consumer debt for individuals, sovereign debt for governments.

South Africa’s Gini coefficient stands at 63.0 the highest income inequality of any country in the world. The top 10% of earners take home 65% of all income.

Source: World Inequality Database

As unemployment rises with 45.6% of young South Africans aged 15–34 now outside employment, education, and training  the consumer base that sustains business growth contracts further. The rational response for a company under margin pressure is to cut costs. And AI hands them the most powerful instrument for doing so in history.

The self-defeating loop

When a company automates to reduce headcount, it improves its own margins. Individually rational. Collectively, when thousands of companies make the same decision simultaneously, the result is a contracting wage base fewer people earning salaries, less money circulating, less demand for the very goods and services those same companies are trying to sell.

The branch-cutter acts in his own interest. The aggregate of a thousand branch-cutters brings down the forest. Each company’s rational decision becomes everyone’s irrational outcome.

Economists have understood this dynamic for nearly a century Keynes called it the paradox of thrift: what is rational for one actor becomes irrational when all actors do it simultaneously. AI is rapidly accelerating all of that, without the century that industrialisation allowed for adjustment.

What the Vatican understood

On 25 May 2026, Pope Leo XIV released Magnifica Humanitas an encyclical on AI, the economy, and human dignity. It was signed on the 135th anniversary of Rerum Novarum, the Church’s 1891 response to the Industrial Revolution. This is the same crisis, a century later.

The message is not anti-technology. It is a reminder that innovation must consider society’s welfare: “The pursuit of greater profits cannot justify choices that systematically sacrifice jobs, because the human person is an end, not a means.” The encyclical calls for social criteria for innovation assessing a technology’s human impact before it is deployed at scale, not after.

The social innovations being proposed and the evidence

The branch does not have to break. If the productivity gains of AI were shared in ways that kept the branch intact, the outcome is different entirely.

Universal Basic Income provides every citizen a guaranteed income floor regardless of employment status. In South Africa, the Social Relief of Distress grant provides a partial precedent. The core argument that automation’s gains should circulate broadly rather than concentrate narrowly is gaining serious traction in global policy conversations.

Reduced working hours distribute available work across more people. The largest controlled study ever conducted on this, tracking 2,896 employees across 141 companies in six countries, published in Nature Human Behaviour in July 2025, found that four-day work weeks improved wellbeing and maintained productivity. 90% of companies retained the arrangement after six months.

Source: Fan, Schor et al., Nature Human Behaviour, July 2025

Cooperative and worker ownership ensures that when automation generates productivity gains, the workers share in them. If the people on the branch also own the tree, they tend to stop cutting. New governance frameworks including mandatory workforce impact assessment before AI deployment create the guardrails that individual companies cannot impose on themselves without collective action

Consider before you cut

AI is genuine. The productivity gains are real. The opportunity for South Africa a country with one of the world’s youngest populations and one of its most urgent talent challenges in the form of AI jobs and workforce transformation is significant.

But technology without people is just automation. Automation without demand is just overhead. The branch-cutter and the tree are the same system. What destroys one eventually destroys the other.

The choice is not between AI and no AI. It is between AI deployed inside a social framework designed for human flourishing, and AI deployed in the absence of one. South Africa cannot afford the second option. The margin for error is too small, the inequality too deep, the unemployment too high.

iFundi does not claim to have resolved these questions. But for 25 years we have been building workforce solutions at the intersection of employer need, learner ambition, and social responsibility. We believe that organisations willing to think about all three simultaneously the technology, the skills, and the social framework are the ones that will build something durable. That conviction is why we are convening this conversation in July.

The question

The technology conversation in South Africa in 2026 is largely about adoption: how fast, which tools, which workflows to automate first. The harder question who benefits, on what terms, and what new social arrangements are needed alongside this future of work is barely on the agenda. It needs to be.

Look in the mirror. Consider what the current trajectory produces. Then decide whether that is the future you want to build.

People Also Ask (FAQ)