Do More With Less

It's Tuesday, March seventeenth, and welcome to HigherIQ. Meta is reportedly planning to cut twenty percent of its workforce, roughly fifteen thousand people, to offset the cost of spending up to a hundred and thirty-five billion dollars on AI infrastructure this year. The stock went up three percent on the news. When Block cut forty percent of its staff in February and credited AI, the stock went up twenty percent. Amazon cut sixteen thousand jobs in January citing automation. The market is sending a very clear signal: fire people, invest in AI, and we'll reward you for it. But there's a company that already ran this experiment to its conclusion, and the results are worth understanding before we assume the bet always pays off. Today we're going to talk about what actually happens when companies replace humans with AI, what happens to the people who stay, and why the most interesting data isn't about the jobs that disappeared. It's about the ones that got harder.

Let's Start with the numbers. So far in 2026, AI has been cited in over twelve thousand job cuts in the United States, according to Challenger, Gray and Christmas, the outplacement firm that tracks this. In 2025, companies directly attributed fifty-five thousand layoffs to AI, more than twelve times the number from two years earlier. The biggest names are familiar: Amazon, Meta, Block, Intel, HP, Workday, Salesforce, Oracle. The pattern is consistent. The company announces layoffs, cites AI as a driver of efficiency, and the stock price goes up. Jack Dorsey's letter to Block shareholders was the most explicit version: "A significantly smaller team, using the tools we're building, can do more and do it better."

That phrase, "do more with less," is worth pausing on. It's not new. In the late nineties and early two thousands, companies adopted enterprise software, systems like SAP, Oracle, and Salesforce, and used the efficiency gains to justify flattening organizations and cutting middle management. The pitch was identical: the technology makes us more efficient, so we need fewer people. What actually happened was that the remaining employees absorbed the administrative work that managers used to do, plus learned the new software, plus kept doing their original jobs. The Nobel Prize-winning economist Robert Solow captured the paradox in a line that's become famous in productivity research: "You can see the computer age everywhere but in the productivity statistics." The technology was real. The efficiency gains were real. But the benefits flowed to shareholders, not to the workers who stayed.

We may be watching the same pattern repeat. Oxford Economics published an analysis earlier this year noting that if AI were truly replacing labor at scale, productivity growth should be accelerating. It isn't. Their conclusion was blunt: AI use in most companies remains "experimental in nature and isn't yet replacing workers on a major scale." They also suggested something uncomfortable: that some companies are using AI as a narrative to dress up cost cuts as forward-looking strategy rather than admitting to past overhiring. Forbes called this "AI washing," and it's become common enough that researchers and analysts now treat it as a distinct phenomenon. The hiring numbers tell the story. Meta had about thirty-six thousand employees in 2019. By late 2022, it had eighty-seven thousand. Zuckerberg himself called the pandemic growth a "permanent acceleration." It wasn't. Even after this proposed twenty-percent cut, Meta would still employ roughly sixty-three thousand people, nearly double its pre-pandemic headcount. Amazon similarly doubled its workforce during COVID. The companies announcing the biggest AI-attributed layoffs are, in many cases, the same ones that hired most aggressively during 2020 to 2022, and the cuts are often bringing headcount back toward pre-pandemic levels, not below them. AI is real. But so is the overcorrection from a hiring spree that turned out to be temporary.

But let's set aside the question of whether every AI layoff is genuine and look at a company that went all in. In 2023, Klarna, the Swedish buy-now-pay-later company, stopped hiring entirely. By 2024, it had partnered with OpenAI, cut seven hundred customer service workers, and replaced them with AI chatbots. CEO Sebastian Siemiatkowski declared publicly that "AI can already do all of the jobs that we, as humans, do." The company reported saving ten million dollars. The chatbot handled two-thirds of all customer interactions. It was the clearest test case in the industry: a major company fully replacing human workers with AI and measuring the results.

By mid-2025, Klarna was rehiring humans. Siemiatkowski admitted the company had gone too far. "We focused too much on efficiency and cost," he told Bloomberg. "The result was lower quality, and that's not sustainable." Customer satisfaction had dropped. The chatbot couldn't handle nuance, empathy, or angry customers dealing with missed payments. It functioned, in the words of one tester, "basically as a filter" that routed people to the human agents who were no longer there. Klarna is now piloting what it calls an "Uber-style" model, hiring remote gig workers to rebuild the customer service function it had automated away. The company's headcount had dropped from over five thousand to about three thousand four hundred during the AI push. It's now adding humans back, but on gig terms: flexible schedules, remote work, no guarantee of hours. The jobs came back. The stability didn't. Forrester Research found that fifty-five percent of employers who made AI-related layoffs now regret the decision. The pattern they documented: companies cut workers based on AI's potential, not its performance, and when the technology underdelivers, they either rehire at lower salaries or fill the gaps with offshore labor.

The piece of this that gets the least attention is what happens to the people who don't get laid off. When a company cuts twenty percent of its workforce and hands the remaining eighty percent AI tools, the workload doesn't shrink by twenty percent. It redistributes. At Block, internal accounts described management warning engineers that productivity targets would increase as AI tools rolled out. Teams that once had eight engineers were reduced to one. Gallup's data shows that U.S. employee engagement fell to a ten-year low at the end of 2024, with only thirty-one percent of workers engaged. Eighty-two percent of white-collar workers across North America, Europe, and Asia reported some level of burnout, according to DHR Global. Forrester identified a growing category of workers it calls "coasters," employees who are disengaged but haven't quit, who don't think their employer deserves their energy. That group is projected to reach twenty-eight percent of the workforce in 2026. The connection to AI layoffs isn't hard to draw: employees watch their colleagues get cut in the name of technology, absorb the extra work, and quietly disengage. There's a well-documented pattern in organizational research that after layoffs, the first people to leave voluntarily are the highest performers, because they have the most options. The people who stay are disproportionately the ones who feel they can't afford to leave. Companies end up with a workforce that's smaller, more burned out, and less capable than the one they started with, which then becomes the justification for more AI investment. The cycle feeds itself.

The deeper question is whether this cycle is different from the enterprise software wave of the two thousands or whether it just moves faster. A recent analysis in The Conversation made a useful distinction between two types of AI-driven workforce reduction. In the first, AI genuinely increases productivity and fewer workers are needed for the same output. In the second, the layoffs aren't a consequence of AI but a way to fund it. Meta is a clear example of the second type: the fifteen thousand workers being cut aren't being replaced by AI today. They're subsidizing the AI infrastructure their employer is betting on for the future. The workers are paying for a product that hasn't arrived yet.

There is a version of this that works. Companies that use AI to handle repetitive, pattern-heavy tasks while keeping humans on the complex, judgment-intensive work are seeing real gains without the backlash. Klarna's revised model, where AI handles straightforward questions and humans handle everything else, is performing better than either approach did alone. But that model requires keeping people, training them, and redesigning workflows around human-AI collaboration. It's slower, less dramatic, and doesn't produce the kind of headline that makes a stock jump twenty percent in a day.

So when someone says AI is taking everyone's jobs, here's the more honest version. Companies are cutting staff and citing AI, but the actual productivity gains haven't shown up yet in the data. The one company that ran the full experiment, replacing human workers entirely with AI, had to reverse course and start rehiring within a year. The stock market rewards layoffs because they reduce costs immediately. But the people who stay absorb the extra work, engagement drops to decade lows, and the best employees, the ones with options, leave first. The pattern isn't new. It happened with enterprise software twenty-five years ago. The technology was real then too. The question was always the same: who captures the gains? Right now, the answer is shareholders. And at companies like Meta, where fifteen thousand people are being cut to fund a hundred-and-thirty-five billion dollars in AI infrastructure, the workers aren't being replaced by AI. They're paying for it.

Stay informed, stay curious, and we'll see you tomorrow.