In September 2025, Salesforce CEO Marc Benioff said the company slashed 4,000 customer service roles, opting for the remaining 5,000 support workers to share their roles with AI agents.

“I need less heads,” Benioff said at the time.

But as more companies adopt agentic AI in hopes of replacing or making human workers more efficient, one top economist has noted that customer service roles—particularly those overseas—are only growing.

Citing data from the IT & Business Process Association of the Philippines, Apollo chief economist Torsten Slok noted in a recent blog post that from 2016 through 2025, call center employment in the Philippines has risen each year, nearly doubling to 2 million over the 10-year span.

He also found that from 2021 to March 2026, unemployment rates in the Philippines have decreased from 9% to about 4%, suggesting AI has not displaced offshore workers. In India, unemployment has remained steady at around 7%. The Philippines dethroned India as the largest call center employer about 15 years ago.

Offshore call center jobs began booming in the late 1990s and early 2000s as a cost-cutting measure. The labor is considerably cheaper overseas than in the U.S., with Filipino call center workers earning wages of 15,000 to over 120,000 Philippine pesos per month, or about $243 to $1,948. In the U.S., the average monthly wage for call center workers is about $2,866, according to Indeed.

But these jobs are also among the most susceptible to AI displacement. The Brookings Institution estimated that 86% of customer service representative tasks had high automation potential.

The apparent contradiction of the job’s potential to be automated alongside rising employment points to a centuries-old economic paradox reflected across labor more broadly, according to Slok.

“This is Jevons paradox in action,” he wrote. “As AI makes call center work cheaper and faster, companies are buying more of it, not less.”

Jevons paradox in the 21st century

Jevons paradox refers to an observation made by English economist William Stanley Jevons in 1865: The invention of the Watt steam engine made coal a more efficient energy source. But instead of coal consumption decreasing because of increased efficiency, consumption skyrocketed as energy became cheaper.

In the age of AI, Jevons paradox has been resurrected and applied to how the technology may be impacting the labor market: While tech leaders like Anthropic CEO Dario Amodei have predicted AI will replace white-collar workers, Slok instead posits that the cost of professional work will fall as AI boosts task efficiency, increasing demand for certain roles and ultimately expanding available jobs. 、

“Lower cost per interaction does not mean fewer interactions,” he said. “It means more customers served, more channels opened and more markets worth reaching. The technology that was supposed to shrink the industry is fueling its expansion.”

This is already being seen in some professions previously predicted to be replaced by AI. A decade ago, “godfather of AI” Geoffrey Hinton declared radiology would one day no longer need to be performed by humans because of its potential to be automated.

Instead, Christoph Herpfer, an economist and business administration professor at the University of Virginia’s Darden School of Business, said the number of radiologists in the U.S. has actually increased by 10% over the past decade.

“We actually have a huge shortage of radiologists. So the exact opposite of this prediction has happened,” he told Fortune.

The future of call centers and automation

Labor economists are already seeing evidence of AI increasing the productivity of call center workers, which could bolster the role’s demand. A 2023 study led by Erik Brynjolfsson, director of the Stanford Digital Economy Lab, found that for more than 5,000 customer support agents, an AI-based conversational assistant tool increased productivity by an average of 14% per hour.

His research from a decade ago similarly found that an early AI feature able to translate listings on eBay increased international exports by 17.5%. Emma Harrington, an economics professor at the University of Virginia, believes this principle could be applied to today’s AI boom.

“What we’re now seeing in the data might be a modern version of the same phenomenon,” she told Fortune. “We can trade labor more easily across countries when language can be traded more seamlessly.”

While some economists expect AI’s strengths to bolster productivity and subsequent employment, others argue the technology’s weaknesses will sustain the call center industry.

Benjamin Shestakofsky, a sociologist and information science professor at Cornell University, told Fortune that AI may not yet be sufficiently trained to navigate the complexities of certain problems. Moreover, humans experience something akin to “AI brain fry” if inundated with an increased case load; AI may be able to increase how many customers agents can serve, but humans still have limitations in how much they are able to process.

In other cases, some companies may explicitly seek out human customer service agents simply because of their brand, according to Shestakofsky. As the world becomes more automated, it’s sometimes a relief to talk to a human on the other line.

“Even if the LLMs get so good that eventually they can deal with those complex issues—which I don’t think they can do now—I would still expect that you would see some companies seeing the value and offering more human interactions that help customers feel seen, feel like they’re more than just a number on a spreadsheet,” he said.