If you’re rolling out AI to do more with fewer people, two groups spent the last two weeks pushing back: one with boos, one with an encyclical.

A record executive told Middle Tennessee State students that AI is “rewriting production as we sit here,” and when they booed, he answered, “Deal with it.” He wasn’t an outlier. A speaker at the University of Central Florida called AI “the next industrial revolution,” and the boos started before she finished.

Then, on May 25, Pope Leo XIV released an encyclical that names the students’ fear directly: workers displaced and discarded as AI takes over their jobs, though it’s not necessarily the technology they’re reacting to.

What the encyclical actually says

Magnifica Humanitas is not a “ban the robots” document; it’s a labor document. Leo XIV deliberately tied it to Rerum Novarum, the 1891 encyclical written during the first Industrial Revolution about the conditions of workers, and he chose the name Leo largely to invoke that lineage. The line that should land for anyone designing reskilling programs is its insistence on protecting the dignity of work by designing systems centered on the person, not only on performance.

What the students are actually booing

It would be easy to read the boos as students who just don’t want to use AI, a generation refusing the tool. However, these graduates started college the semester ChatGPT launched, and most of them have used it the whole way through. They’re not only booing the technology. They’re booing the message that keeps getting bolted onto it, that AI is the thing that makes them optional. A Stanford study last August found that young workers aged 22 to 25 in the most AI-exposed entry-level jobs, software development and customer support among them, saw a 13% drop in employment since ChatGPT launched, a figure the authors have since revised upward to 16%, even as older workers in the same roles held steady or grew. The authors called these workers “canaries in the coal mine.” The door they were told to walk through is measurably narrower than the one their older siblings used. So when a CEO says “AI is the future, master it or fall behind,” they don’t hear encouragement. They hear: you are a cost, and here is the technology that prices you out.

The math of cutting vs. building

If you run an organization, the encyclical might read as nice and beside the point. Your incentive is real: AI lets you produce more while paying fewer people. It’s not that you don’t care about people, it just doesn’t show up in the numbers you report, and a blog telling you to soften your language is asking you to lose more slowly. The case made in numbers, not dignity, runs the other way: replacing people first tends to cost more than it saves, and the evidence keeps stacking up.

Full replacement tends to underdeliver, and 2026 is the year that became visible. The reversal is common enough now that the term “AI hangover” has started showing up for it. Companies that cut deep in 2024 and 2025, betting that software could swallow whole roles, are quietly putting people back. In a February 2026 Careerminds survey of 600 HR leaders who’d run layoffs, two in three had already rehired for jobs they’d eliminated, and nearly a third spent more putting people back than they’d saved by cutting them. For those firms, the cut cost more than it saved. Gartner expects half of all companies that cut customer-service staff citing AI to rehire for the same functions by 2027, and Forrester’s 2026 Future of Work report found 55% of employers already regret their AI-related layoffs. The reasons are consistent: quality slips, customer complaints, and institutional knowledge that walked out the door. AI is excellent at tasks and shaky at whole roles. So map the work at the task level, not the headcount level. Which tasks can AI absorb, and which judgment has to stay human? Keep people on the high-stakes cases, where a confidently wrong answer costs you a client. That’s the work that justifies the salary.

Anyone who’s been stuck in a chatbot loop on an airline’s site at 11pm, just wanting a human who can actually fix their problem, already knows what a “quality slip” feels like from the other side. That’s what whole-role automation produces: not failure exactly, just the steady erosion of the part customers actually valued.

“But my people do the manual stuff AI can actually replace.” Maybe, and where a model genuinely does the whole job better, let it. Nobody’s arguing you keep people doing truly automatable work out of sentiment. But notice how much weight “manual” is carrying. Klarna, the fintech that replaced 700 support agents with an AI assistant, saw roles that looked fully scriptable too. Then the edge-case conversations turned out to be the part that actually mattered. Most “routine” jobs are full of small judgment calls you don’t see — the person who catches the wrong invoice, flags the order that doesn’t seem right, knows which supplier always slips. Automate the role wholesale and you meet those cases the expensive way.

Automating the task and cutting the person are two different decisions, and collapsing them into one is what caps a company. The question isn’t whether AI can take the routine work — often it can. It’s what you do with the capacity that frees up. Shrink into a cheaper version of today’s company, or redeploy that time into work you could never afford to staff before? PwC’s data is blunt about which wins: the most AI-exposed industries are now posting roughly three times the revenue-per-employee growth of the least exposed, and that comes from firms treating AI as a growth engine. Cut-and-shrink locks your ceiling at “same business, fewer people.” Automate-and-redeploy raises it.

A cut also damages the people who stay. After a layoff, the survivors don’t quietly get more productive out of gratitude. In research by leadership-training firm LeadershipIQ, 74% of remaining employees said their own productivity dropped and 77% saw more mistakes around them. Employee-analytics firm Culture Amp found organizational commitment falls about 13.5% after a layoff, and low commitment is one of the strongest predictors of who walks next. So the savings you penciled in are partly fiction: you modeled the salaries you removed, not the output you lost across everyone still on payroll. If you’re going to cut, price that in — the productivity dip, the turnover, the rehiring — because it’s real money, and it lands the quarter after the savings do.

Cutting also starves your own pipeline. The roles that go first are entry-level and back-office — exactly the door the booing graduates can’t find. But those jobs were never just labor; they were how you grew the mid-level and senior people you’ll need in five years. Delete the bottom rung and the ladder collapses behind it on a delay. So defend the bottom rung on purpose: redesign junior roles around AI — people doing higher-judgment work sooner, the tool handling the grind — instead of erasing them. It’s nearly free, and it’s the cheapest succession plan you’ll ever buy.

The return was never in the headcount itself. PwC’s 2025 Global AI Jobs Barometer, built on close to a billion job ads, found workers with AI skills command a 56% wage premium, more than double the year before. That’s the market pricing the scarce thing: not the tool, the judgment to aim it. The companies winning on AI aren’t the ones with the fewest people; they’re the ones whose people are the most leveraged. So treat reskilling as the thing that converts AI spend into AI return — budget it as ROI realization, not HR goodwill. You don’t get the productivity gains by buying the licenses. You get them when people can actually use them. The companies with disappointing AI numbers mostly bought the tool and skipped this step.

None of this means the people betting the other way are naive. Some of the most serious voices in AI, Anthropic’s Dario Amodei among them, expect dramatically smaller workforces, and on a long enough timeline they may be right. But “right eventually” and “right now” are different bets. The companies showing up in the rehiring numbers weren’t wrong about where AI is going; they were wrong about when it got there. Cutting before you’ve remapped the work doesn’t put you ahead of the curve; it puts you in the Careerminds survey.

The through-line

It might seem like a pope and a stadium full of 22-year-olds would rarely agree on much. But they agree on this: the deepest objection isn’t only to the technology. It’s to being treated as the easiest thing to cut. What both are really defending is dignity, the idea that a person is worth building around, not a cost to be subtracted.