South Africa’s structural challenges run deeper than those of most economies. We face a multi-generational unemployment crisis in one of the most unequal and poverty-stricken countries in the world.
Treating AI as nothing more than a cost reducer will not fly for much longer. If we are to make any dent in unemployment, poverty and inequality, business, government and civil society must ask how AI can increase productive capacity, create new forms of work and widen access to digital skills. The decisions we take now will be judged by future generations, not on how tightly we managed payrolls, but on whether this technological shift meaningfully expanded participation in our digital economy.
A recent OECD report, “How do people experience new technologies and generative AI?”, covering 14 countries including South Africa, is instructive on where we find ourselves and where the opportunities lie.
The data shows that AI training is being taken up mostly by younger, more highly educated people, which the report attributes to an awareness of career shifts. Some 46% of professionals aged 26-35 undertook formal AI training over the past year, against 39% of those aged 18-25, 38% aged 36-45 and fewer than 20% of workers over 55. Tertiary-educated employees were twice as likely to seek out AI learning as those with only a secondary education.
This points to a growing risk: AI futureproofing is becoming self-selecting, driven by younger, more highly educated employees, while older, experienced or less educated workers slip away.
The deeper issue is not that older professionals appear less interested in technology than younger peers. It is that South African employers, in both the private and public sectors, are at risk of allowing AI capability to become a voluntary opt-in even as fluency becomes a baseline expectation across white-collar roles.
Slipping behind
The people who already have digital confidence, fast connectivity, spare time and natural curiosity are the ones running experiments and introducing new solutions. Younger, more digitally fluent and better-educated workers are streaking ahead, while the irreplaceable institutional knowledge of senior leadership fails to keep pace. And the majority of our youth, who are not tertiary educated, are slipping further behind.
AI adoption therefore demands a serious change-management rethink. Older, more experienced workers are more likely to engage with AI when it is positioned as a way to extend their expertise, cut administrative drag and streamline knowledge transfer rather than as a signal that their experience is suddenly obsolete. Younger, less educated workers, in turn, need to be brought into the opportunities AI has opened up, in ways that matter for their economic prospects.
But the disconnect goes deeper than training metrics.
The report also reveals a growing trust deficit. More than 75% of under-35s view AI as useful, and trust and ethical confidence in it both decline with age. Older users remain overwhelmingly suspicious or hesitant or answered with a flat “don’t know”.
In a multi-generational workplace, younger employees happily hand cognitive tasks to generative AI tools they implicitly trust, while their managers view the same tools with scepticism. That can quickly turn into conflict: juniors move fast with AI-generated summaries or analysis, while senior managers question the integrity, confidentiality and reliability of the output.
Both sides have a point. Blind trust is risky, exposing a business to algorithmic bias, data leaks, hallucinations and liability. But wholesale rejection paralyses productivity and chokes innovation.
Read: AI will leave the world short of workers, says Jeff Bezos
To counter both the self-selection trap and the trust divide, workplaces must shift their learning models, moving AI training from a voluntary benefit to a structured workforce-capability programme. In practice that means role-based learning pathways. A financial analyst, call-centre manager, HR business partner, lawyer and executive assistant do not have the same training needs. Each must understand how these tools change their work, where they introduce risk, where they require human judgment and how they enhance their particular value.
The report also places South Africa among the countries with the highest generative AI uptake and the most positive perceptions, especially among the young. That organic optimism is a real macroeconomic asset. But optimism alone will not create inclusion. Without deliberate intervention, AI can deepen inequality, because those most likely to benefit are already educated, connected and digitally confident.
There are three ways business and government can harness this enthusiasm and counter the risk of creating another layer of digital exclusion:
- Create entry-level AI-enabled roles. Rather than letting AI erase the bottom rungs of the corporate ladder, we must deliberately design junior roles that did not exist two years ago: data-cleaning assistants, prompt-library coordinators, customer-insight analysts, AI quality reviewers, automation support staff, knowledge-base curators and digital-workflow assistants.
- Augment labour-heavy sectors. The real multiplier effects come when AI moves beyond head office. There is untapped potential in deploying generative AI to lift productivity and cut administrative burdens across agriculture, healthcare administration, education support, small-business services, logistics and township enterprise.
- Invest in community-based and work-integrated learning. Focused, sustainable partnerships with universities, TVET colleges, tech bootcamps and youth employment programmes can put practical AI skills within reach across both urban and rural areas.
Real leadership now requires deliberate alignment between the promise of technology investment and our economic and employment realities. We have abundant opportunity to automate mundane tasks, not human potential. We should seize it, for the generations to come.