Something shifted at the IMF and World Bank annual meetings in Bangkok this week. For years, policymakers argued about whether artificial intelligence would really disrupt jobs. On Monday they stopped arguing about whether and started arguing about how to cushion the fall. Managing Director Kristalina Georgieva described AI as a "slow-moving tsunami," and her estimate is unchanged but newly urgent: about 40 percent of jobs worldwide will eventually be augmented, downgraded, or eliminated, rising to 60 percent in advanced economies.
The word worth dwelling on is "slow." A fast shock forces panic; a slow one leaves time to decide, seat by seat, what each job becomes. But slowness is cold comfort to the group already feeling it. Georgieva was blunt about who gets knocked down first: "young people that are coming into the labor market." They used to take routine, repetitive jobs and climb from there. "But if they're gone," she said, "I don't think any young person wants to be a barista in a coffee shop as a professional future."
Singapore's President Tharman Shanmugaratnam, who co-chairs the World Bank's Jobs Council, sharpened the point: "the first rung in employment, your first job upon graduation, is now seeing less demand." In the United States, he noted, demand for graduates has fallen for the first time in a century, and the wage premium for a degree is shrinking. The danger is not only fewer entry roles but the loss of the work that once trained future seniors. Strip the learning tasks out of a junior job and you have not just made it smaller; you have removed the path by which this decade's graduates were meant to become the next decade's experts.
Set against this is a stubbornly cheerful macro story, and the tension between the two is the real news. The same AI boom that is thinning entry-level hiring is also propping up the world economy, keeping global growth near 3 percent despite an energy shock. Georgieva argues that if advanced Asian economies capture the productivity gains, the lift could more than offset the drag of their aging, shrinking populations.
Not everyone buys the arithmetic. The economist Paul Krugman, writing the same week, points out that a typical AI data centre costs billions yet employs between 50 and 150 people once running. Spend eight billion dollars almost anywhere else and you support more than ten thousand jobs; spend it on a data centre and you support around a hundred. His back-of-the-envelope conclusion is that the capital flooding into AI will employ roughly ten million fewer workers than the same money invested elsewhere, pushing down wages and tilting income further from labour toward capital.
There is a more hopeful ledger too. New McKinsey research projects the US economy will have more jobs in 2035 than today, but with a churn that forces some 11 million Americans out of shrinking occupations and into new ones, at more than triple the historical rate of cross-occupation moves. The catch is the hurdles: credentials required for 85 percent of growing jobs, and months of forgone wages that a worker with a child cannot simply absorb. Every forecast in Bangkok, optimistic or grim, circled the same point. The wave is coming slowly enough to prepare for, which is precisely why doing nothing is the one response guaranteed to fail.