The Jobs AI Is Actually Replacing (And the Ones It Isn't)
2026-03-05
75% of programming tasks now covered by AI — yet no measurable increase in unemployment for exposed workers. The gap between what AI could theoretically do and what it's actually doing is the most important chart in AI economics right now.
AI isn't eliminating jobs. It's rerouting who gets hired into them. The headline — no systemic unemployment increase — will comfort optimists. It shouldn't. Hiring of 22–25 year olds into exposed occupations has dropped 14%. AI isn't firing people. It's closing the door for the next generation walking in. For the Gulf, every nationalisation strategy depends on young citizens entering knowledge-economy roles — precisely the most exposed. For Africa, 60% of the continent is under 25, educated for BPO, data processing, and customer service — the top AI-exposed occupations. This is not a labour market paper. It is an early warning system for every economy betting its future on knowledge workers.
The most exposed workers earn 47% more, are older, more educated, and female. Stop planning for AI as a factory-floor disruption. Start asking what happens when the knowledge-economy jobs you've been training an entire generation for start absorbing fewer graduates. AI displacement isn't coming for low-wage workers first — it's coming for your most expensive, most educated employees.
Anthropic measured what AI could theoretically do to jobs — then checked what it’s actually doing.
The gap between those two things is the most important chart in AI economics right now.
What the data shows
75% of programming tasks are now within AI’s observed coverage. Customer service representatives: ~70%. Data entry: 67%. These are the headline numbers that will circulate.
But here’s what most commentators will miss: there is no measurable increase in unemployment for highly exposed workers since late 2022.
Not because the exposure isn’t real. Because the adoption hasn’t fully arrived yet.
The number that should concern you
Hiring of 22–25 year olds into exposed occupations has dropped 14%.
AI isn’t firing people. It’s closing the door for the next generation walking in. The experienced workers keep their jobs. The young workers stop getting hired into them.
Why this hits differently in the Gulf and Africa
Every Gulf nationalisation strategy — Saudisation, Emiratisation, Qatarisation — depends on young citizens entering knowledge-economy roles. Those are precisely the roles most exposed.
For Africa, the stakes are sharper. 60% of the continent is under 25. Educated for BPO, data processing, customer service — the top three AI-exposed occupation categories. The demographic dividend becomes a liability if those entry-level doors close before young workers can walk through them.
The counterintuitive finding
The most exposed workers earn 47% more than average. They are older, more educated, and disproportionately female.
AI displacement isn’t coming for low-wage workers first. It’s coming for your highest payroll line items.
“We find no systematic increase in unemployment… though we find suggestive evidence that hiring of younger workers has slowed.”