AI is creating jobs, but SA could still lose
The artificial intelligence (AI) revolution was supposed to empty offices. For now, it is also filling construction sites. Somewhere between these two images lies a question that South Africa (SA) cannot afford to ignore: Will we capture the new opportunities, or mostly experience disruption?
The Economist estimates that AI has created roughly one million jobs in the United States, compared with about 200 000 retrenchments attributed to AI since mid-2023. The gains include software specialists, but also the electricians, technicians, and engineers needed to build and power data centres. Intelligence may be becoming artificial, but the infrastructure behind it remains stubbornly physical. These figures, however, deserve caution. Not every additional engineering job owes its existence to AI, and counting the announced retrenchments misses people who were never hired. Construction booms also end. The early evidence challenges predictions of immediate mass unemployment. It does not settle what happens once the infrastructure is built and the technology matures.
Nevertheless, the economic opportunity extends well beyond construction. Making a service cheaper can create customers who previously could not afford it. Imagine a small accounting practice using AI to process routine paperwork. It could dismiss staff and keep serving the same number of clients. Or, it could lower fees, reach hundreds of smaller businesses, and employ more people to advise them. More output does not automatically mean more employment: Demand must grow enough to offset the labour saved on each task. Competition matters here. If productivity gains translate into lower prices, they can widen access. If they remain mostly higher margins, the employment benefits may be narrower.
For SA, this distinction is uncomfortable. We could lose routine administrative work here while the investment, technical jobs, and profits accumulate elsewhere. A global jobs boom offers little comfort to someone retrenched in Johannesburg if the replacement vacancy is for an electrical engineer in Indiana. Nor does that worker become an engineer merely because an economist calls for “reskilling”. Training takes time, money, and credible pathways into employment. People have bills to pay while they learn. A transition can be positive for the economy in aggregate, but brutal for the households caught between occupations.
There is also a less obvious danger within companies. Junior employees, traditionally, learn by doing the routine work that senior staff no longer need to do. If AI takes over that work, how do tomorrow’s experts acquire judgement? Saving on graduate recruitment today could leave firms with a skills shortage of their own making later. Employers, therefore, need to redesign entry-level development, giving young staff supervised responsibility for checking AI output and solving client problems. Knowing how to prompt a chatbot is useful. Knowing when its confident answer is wrong is considerably more valuable.
Government’s contribution is equally practical: Dependable electricity, affordable connectivity, and training linked to actual employer demand. But, attracting data centres alone is not an employment strategy. Their construction creates work; sustained operating jobs are a different matter. The bigger opportunity may lie in helping thousands of existing businesses become productive enough to grow. We need not invent the next global AI model to benefit from it. An accounting firm serving businesses that previously could not afford advice is also part of this revolution.
Investors should resist a different temptation: Confusing an economic breakthrough with a guaranteed investment return. AI can transform businesses while particular shares disappoint because their prices already assume extraordinary success. Equally, an infrastructure spending boom can strain electricity supply and financing before its productivity benefits reduce costs. Technological progress does not arrive with permanently cheap money attached to it.
The useful question, then, is not whether AI creates or destroys jobs. It does both. SA’s challenge is to turn cheaper expertise into more customers, stronger businesses, and routes into skilled work. If our firms use AI only to shrink their wage bills, we may become more efficient without becoming more prosperous.
This article has been published on Moneyweb.
