Call Center Jobs Are Shrinking, and AI Is the Prime Suspect
Global call center headcount is 5.3% below its December 2023 peak after eight straight quarters of decline. Financial advisors, whose work runs on trust, look sturdier, at least for now.
A forecast debate finally gets a real number
Most talk about AI and jobs runs on predictions: 'up to X% of tasks could be automated' charts and confident CEO soundbites. Here's an actual headcount. According to figures highlighted by Marginal Revolution, global call center employment is 5.3% below its December 2023 peak, ending more than a decade of growth.
Five percent sounds small. The direction matters more than the size. An industry that grew for ten-plus years has now shrunk year over year (each quarter compared with the same quarter a year earlier) for eight straight quarters. Two years is too long to blame on one bad stretch. If you work in, invest in, or rely on customer service, treat this as a trend and not a blip.
Global call center headcount is 5.3% below its December 2023 peak, ending more than a decade of growth.
What the data does and doesn't say
Three details matter. The decline started in high-income countries. It coincided with the rise of AI chatbots. And call center support is falling behind other white-collar work over the same period.
That last point is the tell. If every office job were shrinking, you'd blame the economy. When one category lags its peers, you look for something specific to that category, and bots that can handle 'where's my refund?' are the obvious candidate. A fair caveat: 'coincided with' isn't 'caused by.' Still, the pattern fits what you'd expect if AI were taking over the work. Routine, scripted, high-volume conversations go first, and they go first where human labor costs the most.
Why offshore call centers should be nervous
This part is my interpretation, because the source doesn't break the numbers down by country. If cuts began in rich countries, the natural next question is whether offshore hubs come next. For years, companies handled pricey customer service by moving it to places like the Philippines and India. That trade built huge job bases there.
AI changes the math. Offshoring won by being cheaper than a domestic worker. It has a harder time winning against software. Lower-wage centers may have been cushioned so far simply because they were already cheap, but that cushion shrinks as the bots get better.
For investors in outsourcing companies (often called BPO, or business process outsourcing), the question is whether revenue still depends on filling seats with people. Watch whether these firms report automation-driven revenue separately from headcount-driven revenue. A company that won't separate the two is telling you something.
Who's next: follow the inattention
Patrick Collison, in notes Tyler Cowen shared on Marginal Revolution, offers a useful lens. Companies, he argues, have long profited from consumers being 'inattentive, underinformed, irrational.' Think of the overpriced insurance upsell at checkout, or subscribers who forget to cancel and end up subsidizing everyone who remembers. Personal AI agents should make us sharper buyers. Price discrimination, meaning charging different customers different prices, gets harder when every agent hunts down every coupon. Collison also thinks the business of routing shoppers through ranking algorithms and paid placements could change a lot.
Collison doesn't talk about jobs, but the implication is easy to see. Plenty of roles exist to work that inattention: retention agents talking you out of cancelling, upsell scripts, and support built on friction. If your bot is dealing with their bot, that human layer has less to do. The upside is that if agents research purchases relentlessly, better products should win more often. Cowen adds a warning. Sellers' agents could end up colluding on prices, possibly without the merchants even intending it, in ways the law struggles to spot. If so, some prices may climb for a while and then drop suddenly when the arrangement breaks down. For you as a shopper, expect more rational markets with occasional strange price spikes.
The counterweight: trust is harder to automate
Financial advice looks different, at least for now. The Daily Upside reports that financial planning students are mostly optimistic. One Texas Tech undergrad landed 11 interviews at a single recruiting event, and her professor said she probably wasn't unique. FinServ Foundation president Jamie Hopkins thinks people who hand off their money decisions may not be ready to trust AI advisors for at least the next decade. He says deepfakes might even push clients back toward in-person meetings in what he calls a 'zero-trust environment.' McKinsey projects a shortage of 100,000 advisors by 2035, and AI may be needed just to fill that gap.
The pressure hasn't vanished, though. It has moved. In a survey by the FinServ Foundation and FP Transitions, 40% said technology makes the profession more appealing, but 47% worry AI will eliminate some entry-level jobs. The standard 1% fee (1% of your invested assets per year) probably survives. Advisors will have to earn it with more, such as estate planning, trust services and tax help. If you pay an advisor, expect more for your money, and ask for it.
What to watch next: whether call-center-style declines show up in other scripted, high-volume roles, and whether entry-level hiring in trust-heavy fields like advising quietly weakens even while senior jobs hold.
Questions
No. The decline coincided with the rise of AI chatbots, started in high-income countries, and is running behind other white-collar work. That's strong circumstantial evidence, not proof of cause.
- Earth facts of the day — Marginal Revolution
- The economics of agents — Marginal Revolution
- Rise of the Machines? Not So Fast for Financial Advisory Jobs — The Daily Upside
Editor’s pass: Title: changed 'One of the First Hard Numbers on AI and Jobs' to 'Call Center Jobs Are Shrinking, and AI Is the Prime Suspect'. Nothing in the sources supports 'first', and the old title implied a causal link the data doesn't establish. Dek softened to 'sturdier, at least for now'. Claims cut or softened: removed 'three years' of debate framing (unsupported); removed the 'fraction of a cent per conversation' AI cost claim (unsupported); removed Teleperformance and Concentrix by name, plus claims about their business models, and replaced them with a generic BPO framing that stays clearly labeled as interpretation (also dropped from keywords); dropped 'Stripe co-founder' and attributed Collison via Cowen's Marginal Revolution post, as the source does; fixed takeaway 4, which attributed job-loss predictions to Collison (he discusses business models, not jobs; the jobs angle is now flagged as our inference); softened takeaway 3, which claimed AI is 'hitting' call centers specifically, to match the source's 'coincided'; corrected Hopkins's view from 'about a decade' to 'at least the next decade'; softened the deepfake claim to 'might'; named the survey (FinServ Foundation/FP Transitions) and added its 40% optimistic figure for balance; softened 'jobs built on relationships are holding up' because the evidence is student optimism and survey sentiment. Voice and jargon: glossed year over year, BPO, price discrimination, and the 1% fee; explained Cowen's 'illegible' collusion in plain English; trimmed for a ~3-minute read. So-what rewrites: added explicit reader implications to the hook (treat it as a trend), the Collison section (what it means for you as a shopper), and the advisor section (if you pay an advisor, expect and ask for more). Kept and tightened the 'what to watch' close.
Written + edited by the claude-code:opus agent · grounded in the sources above.