South Africa cannot afford an AI-ready generation with nowhere to start

South Africa does not have the luxury of treating the disappearance of entry-level work as a minor side effect of artificial intelligence. The country already has a first-job crisis.

Gleb Tsipursky, PhD, is a behavioural scientist, CEO of Disaster Avoidance Experts, and author of The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026).

Statistics South Africa reported that unemployment among people aged 15 to 24 reached 60.9% in the first quarter of 2026, while unemployment among those aged 25 to 34 stood at 40.6%. Millions of young people are trying to enter a labour market that already demands experience before many of them have had a chance to gain it.

The country is also investing in future skills. In March, the Department of Higher Education and Training announced a partnership with Google that includes 5,000 Google Career Certificate scholarships in areas such as AI Essentials, cybersecurity, and data analytics, with a focus on public universities, TVET colleges, CET colleges, rural areas and townships.

That is useful. But training creates value only when people can convert it into work. President Cyril Ramaphosa made that point directly on Youth Day, highlighting the Jobs Boost Outcomes Fund , which enrolled more than 9,000 young people and placed more than 7,000 into employment. He argued for moving away from training for training’s sake and called on employers to hire for potential rather than requiring experience before a young person has had a first chance.

AI makes that challenge more urgent. The Stanford Digital Economy Lab’s Canaries Dashboard shows the strongest employment weakness in highly AI-exposed occupations among workers ages 22 to 25. The dashboard also finds a clearer negative relationship where AI use is oriented toward automation rather than human augmentation. Stanford cautions that the US payroll sample is not nationally representative and does not prove AI caused the changes. South Africa should still treat it as an early warning.

The wrong response would be to preserve routine tasks simply because junior workers used to do them. If AI can draft a standard report, reconcile routine information, or generate a first version of code, firms should use the efficiency. The better question is what junior workers do next.

A beginning analyst can use AI for the first pass and then learn to test data, investigate exceptions, and defend conclusions. A junior customer service employee can let AI handle routine responses while taking more complex cases under supervision. A new developer can use generated code while being judged on testing, security, and maintainability.

South Africa should bring the Jobs Boost logic into AI adoption. Public incentives, training partnerships and large employer AI programs should track paid junior placements and sustained progression, not only course completions or productivity gains. Employers should be able to explain which developmental tasks replace the ones they automate.

The country is spending scarce resources to equip young people with digital skills. It should protect the bridge from those skills to experience.