There is a question that has been lurking beneath the surface of every breathless announcement about AI productivity gains, and it is one that business has been reluctant to pose directly: if AI can do the work of three developers, what happens to the other two?
Companies from Jack Dorsey’s Block to tech giant IBM and enterprise software provider Salesforce have already collectively cut tens of thousands of jobs in the name of AI efficiency.
The technology sector, for decades a reliable generator of skilled, well-paid employment, is rapidly automating itself. And South Africa — with its 32% unemployment rate and its fragile, skills-dependent economy — has its head buried in the sand.
Morgan Goddard, a partner at Reunert subsidiary iqbusiness, offered me a candid window into the transformation under way in software development. The shift now taking place, he said in a recent interview, is not incremental. Agentic coding assistants now autocomplete, suggest, test and architect software code at speeds that make conventional programming look antiquated.
“These tools are speeding up your coding ability by something like 10,000 times,” he said. “And then, in the past few months, we’ve seen multi-agent orchestration come into the picture — an architect agent, a tester, a designer, a reviewer, a coder — all working simultaneously, autonomously, on your behalf.”
The consequence for team sizes in software development is direct and large. “What used to be a 10-person job is now potentially a two- or three-person job, or even a single person’s job,” Goddard said. That is not a prediction about the distant future. It is a description of what is happening now, in Joburg and Cape Town, as much as in San Francisco and London.
The rise of so-called vibe coding — using natural language to generate functional software without writing a single line of code — has made the disruption more visible. AI developer tools now allow a nontechnical person to describe what they want in a prompt and receive a working application, often within hours.
But there is a more profound problem buried in the productivity story, one Goddard identified and that deserves much more attention in South Africa. Senior developers are battle-tested precisely because they spent years writing code, making mistakes and developing an intuitive feel for their craft. If junior developers never acquire that grounding because AI agents are doing the foundational work from the outset, where do tomorrow’s senior developers come from?
“Do we still train people in syntax and the fundamentals before we allow them an agent?” Goddard asked. “Or do we just not care, because the agents are getting better and better?”
It is a question without an answer — yet. South Africa has invested significantly, if unevenly, in building a pipeline of software engineers. Universities, coding bootcamps and other digital skills programmes have all been premised on the assumption that entry-level developer jobs would exist to absorb graduates. That assumption is now in doubt.
And the AI jobs reckoning is not confined to software development. Across the economy, the knowledge worker roles governments and economists once identified as the answer to structural unemployment — from data processing to financial analysis and customer support to content moderation — are precisely those most exposed to AI substitution.
When Dorsey said recently that a significantly smaller team using AI tools could do more and do it better, he was not making a futuristic claim. Block recently laid off 4,000 employees, almost half its workforce, with Dorsey pointing to AI as the reason.
What makes South Africa’s exposure distinctive is the timing. The country never fully industrialised, never built the manufacturing base that absorbed low-skilled labour in East Asia, and has staked much of its economic strategy on growth in services. The tech sector was supposed to be part of that story.
The uncomfortable truth is that AI is not arriving gradually enough for society to adapt through normal processes. As Goddard put it, every other revolution happened at a pace society could cope with. This one is very different.
South Africa needs an honest national conversation about what AI means for employment, skills development and economic inclusion. I’m not talking about the cheerful, conference-circuit version about opportunity and upskilling South Africa is so good at hosting, but a harder reckoning with the structural risks. AI is rapidly changing the world of work, and South Africa is wholly unprepared for the tsunami that’s coming.
McLeod is editor of TechCentral