No one knows for certain if AI will trigger a white-collar jobpocalypse. The loudest warnings come from those building and selling the technology, blurring the line between prediction and hype. Current data gaps leave everyone guessing, but one script is already emerging: retraining, upskilling, and the hollow language used to make displacement politically palatable.
Policymakers and tech leaders should not wait for mass disruption. America ran this experiment after deindustrialization, and the payoff from retraining was weak at best. The optimistic case is that AI will create new jobs we cannot yet imagine, but as Brookings Institution Senior Fellow Molly Kinder warns, the “messy middle” between today’s rocky adoption and the promised abundance could provoke public revolt.
Bloomberg Economics estimates that 27% of workers in advanced economies-more than 120 million people-could be meaningfully affected by AI. Nearly a quarter of CEOs surveyed say over half their workforce will need upskilling. Yet the definition of retraining remains maddeningly vague. Inside the tech industry, answers like “universal basic income” or “learn a trade” ring hollow, especially as software engineering itself becomes one of the most exposed professions.
A more promising path comes from University of Chicago economist Alex Imas, who asks: “What will be scarce?” If AI makes cognitive tasks cheap, human relationships become more valuable. Education, eldercare, and social work-the “relational sector”-could command a premium, but only if policymakers invest now in funding, apprenticeships, and career ladders. Workers are not anti-technology; they want a say in how it is used. Real-time data tracking and targeted public investments are needed more than comforting slogans.
The risk is that “reskilling” becomes the excuse making mass unemployment politically palatable-a signpost at the edge of a cliff rather than a bridge to an AI future.