Torsten Slok, chief economist at Apollo, published a note arguing that the AI boom has not displaced offshored white-collar work and is in fact expanding it. Fortune wrote it up on 29 August 2026 at 13:08:48 UTC. The argument is a serious one, and its evidence base has a specific edge worth marking.
What the note says
Citing the IT & Business Process Association of the Philippines, Slok notes that from 2016 through 2025 Philippine call-centre employment rose every year, nearly doubling to 2 million. He adds that from 2021 to July 2026, unemployment fell from 9% to about 5% in the Philippines and from around 7% to 6% in India. "If AI were displacing white-collar work at scale, you would expect to see it first in the Philippines and India," he wrote. "Instead, the unemployment rate in both countries has continued to trend lower."
Where the evidence stops
The load-bearing series — the one that actually measures call-centre jobs — ends in 2025. The agentic deployments the argument is about are a 2026 phenomenon. The only 2026 data point offered is a national unemployment rate, and a national rate cannot isolate a sector: business-process outsourcing is a large share of Philippine employment but nowhere near all of it, so a falling headline rate is fully consistent with BPO seats being cut while other sectors add more. The claim may well be right. It is not yet measured for the period under dispute.
A number that gets read backwards
The counterweight cited is a Brookings Institution estimate that 86% of customer-service-representative tasks have high automation potential. That is a task-exposure estimate — the share of activities a model could in principle perform — and it is routinely rendered as "86% of call-centre jobs can be replaced". Those are different statements, and the gap between them is precisely the thing in dispute.
The mechanism, and what it is
Slok's explanation is Jevons paradox: "As AI makes call center work cheaper and faster, companies are buying more of it, not less," and "lower cost per interaction does not mean fewer interactions." That is a coherent account of how a cost collapse can grow a market rather than shrink it — the pattern Jevons described for coal in 1865. But it is a hypothesis offered to explain data that does not yet cover the period, not a finding drawn from it. The supporting analogies point the same way: the radiology comparison invoked in the piece, that US radiologist numbers rose 10% over the past decade against predictions of automation, is a ten-year trend, not evidence about the current cycle.