It’s a shocking reversal: A third of managers say they’ve rehired the human workers AI was going
to replace. Will the about-face last?
Help wanted: Humans
Scroll to read
VIEW THE
PODCAST
READ THE
FULL MAGAZINE
hen one of the copywriters left his marketing agency, Cubic Promote CEO Charles Liu wasn’t worried.
He was excited, even. Presentations, emails, marketing copy: It was all stuff that large-language AI models were
supposed to handle with ease. So instead of replacing the employee, the firm started using one of the world’s
leading AI products, Claude, to do its work.
Claude certainly churned out drafts quickly—far faster than any human could. Indeed, its output enabled Cubic
Promote, which sells more than 15,000 custom-branded products to companies across Australia, to expand its own business. But that, Liu discovered, was also becoming a problem. The reams of content the AI model was producing needed to be edited, fact-checked, formatted, and prepped before any of it could be shown to other people at Cubic Promote, let alone customers or vendors. “Automating those later stages was much more complex than generating the drafts themselves,” Liu says.
Before long, Liu found himself hiring someone who could not only edit the AI content but also assist with the other tasks that transformed an output from interesting raw material into a client-facing finished product. AI is effective, he says, but it needed human help to solve his organization’s problems. “It helps the business grow faster but also creates a need for people who can manage, improve, and complete the work it begins.”
Firms haven’t figured out the right balance between AI and humans.
The
Problem
Billions of dollars and millions
of jobs are at stake.
Why it
Matters
Rethink strategies that assume
AI can automate everything.
The
Solution
W
”
”
Of course, nobody is guaranteeing any favoring of human over machine will last—not by a long shot. Executives, for one, are loath to admit that their AI plunge was a mistake. But the rapid adoption of AI
has certainly been a costly decision, with non-technology organizations on track to spend $280 billion this year alone. Many firms redirected money to AI that had been earmarked for salaries, onboarding,
and other people costs. Now they’re faced with spending even more money on those exact things. Aaron Strout, author of the book Wired for Purpose: Why Humanity Is the Biggest Differentiator in a Digital World, calls this phenomenon “the AI boomerang effect,” and he believes it’s not a fluke or a correction. “It’s the market rediscovering something that was true all along: The future belongs to the most human among us,” he says.
”
that was true all along.”
By Russell Pearlman /
Photo Illustrations by Tim Ames
AI vs. human: A scoreboard
More than three-quarters of US firms have implemented AI in some way. But where
has AI outdone humans, and where has the human touch proved more valuable?
AI BETTER
HUMANS BETTER
Drafting presentations.
Analyzing and interpreting market research.
Understanding client sentiment.
Building client relationships.
REAL ESTATE
Quietly, a bunch of firms that carried out AI-driven layoffs have realized that they still need people. A car manufacturer rehired hundreds of engineers to work on quality-control issues. A big technology firm is now bringing some of its HR professionals back because AI can’t handle the many ethical dilemmas their jobs entail. Indeed, 39 percent of business leaders said they made employees redundant as a result of deploying AI, but more than half of those leaders, 55 percent, said some of those decisions turned out to be wrong, according to a 2025 survey by software firm Orgvue.
Humanity makes a comeback
By the millions, workers have been worried stiff that AI will replace them. After all, not long after the technology’s coming-out party—the public debut of Open AI’s ChatGPT—economists and tech pros alike were saying the technology would soon eliminate the need for 10, 20, even 40 percent of human employees. Today, business leaders seem eager to use the tech wherever they can, because labor is often an organization’s biggest business expense. Plus, even the best human employees are flawed (we can be slow, or stubborn, or demand things like time off). Why pay 100 software engineers when AI can write and test code faster than all of them put together? Who needs 200 salespeople when AI can make all the cold calls and answer client questions? Plenty of organizations, big and small, rushed to implement AI, then slashed jobs by the thousands—or didn’t hire for open roles—certain that the technology would make their firms far more productive.
Catanzaro, vice president of applied deep learning research at NVIDIA, the major supplier of microchips
for AI use.
But the biggest reason may be that leaders, pulled in
by the hype, overestimated how much work AI could actually do. AI does lots of things pretty well, but it needs a human to bring its output—whether it’s a software program or a marketing presentation—to the finish line. Many leaders assumed that since AI absorbs tasks, it actually eliminates jobs, Strout, the author, says. That’s why many companies stopped hiring entry-level employees: They thought AI agents could do all the basic collating, data analysis, website building, and coding often expected of young people entering the workforce. But give AI a situation that requires judgment, negotiation, or relationship building, and it sputters. Even junior employees often perform better in those situations than AI does. “There are still moments when people prefer to speak to another human,” says Petr Chocholka, director of IP development for the Korn Ferry Institute. It’s not because the AI has failed, he says; it’s because trust, familiarity, and accountability feel different when another person is involved.
”
can’t do.”
IMPROVE AI... AND PEOPLE
Some leaders think AI will solve all their problems, and some don’t want to touch it.
Experts suggest leaders make these adjustments to balance AI and human workers:
Redesign roles with AI in mind.
Instead of just assuming AI will cut head count, think about which aspects of a role AI can take over to free up an employee for higher-value tasks.
Check AI’s work.
Never let AI-created products
go out the door without being reviewed by a human employee.
Use AI to speed
up “invisible” assignments.
Focus first on tasks that require minimal human interaction, such as retrieving information or processing transactions.
Give people freedom to experiment.
Encourage and reward individuals and teams for finding creative ways to use AI and sharing best practices across the organization.
FINANCE
SALES
LEGAL
MARKETING
Parsing documents.
Speeding up brief writing.
Arguing in court.
Creating novel legal arguments.
Building account plans.
Maintaining communication with clients.
Closing tough sales.
Reading emotional responses of customers.
Recognizing patterns within investing. Flagging potential fraud.
Making high-stakes capital-allocation decisions.
Managing crises.
Staging virtual home tours.
Targeting potential buyers with properties.
Interpreting homeowner-association documents. Understanding local development issues.
But in a rare reversal, companies are quietly backpedaling on many aspects of the AI revolution. Most importantly, they’re rehiring many members of the human species because of their real human traits. The total number is hard to know, but a 2026 survey of 2,000 hiring managers found that 32 percent—almost a third—had eliminated a role primarily due to AI, only to be forced to rehire someone for the same or similar position. In some cases, it’s a cost issue—AI has become far more expensive since the major platforms changed their pricing models this spring. Some tasks firms once delegated entirely to AI are now far less expensive if humans do them. Other leaders found that they miscalculated how much work they could automate. But in most cases, leaders are finding that humans have skills, experiences, and decision-making abilities that AI can’t replicate—at least, not yet.
Firms are learning that the AI hype—the idea that the technology would allow firms to cut their workforces immediately—was misconceived from the start, says Bryan Ackermann, Korn Ferry’s head of AI strategy and transformation. “We’re getting our first experience, at scale, of what AI can and can’t do,” he says.
Experts point to a couple of reasons why organizations might have miscalculated. The first was somewhat out of their hands. After many firms spent large sums to
get their back offices and data computers ready to implement the technology, the big AI firms that hadn’t been charging for their tools dramatically raised their prices by 20 percent to 40 percent. (The firms had burned through so much cash.) That caught everyone off guard. Suddenly, as fees matched how much work the AI agents performed, AI-automated workflows
were no longer cheaper, and in some cases, they were far more expensive than human employees had been. Even those who have the most to gain from the AI revolution say the changing costs of the technology
are an issue. “For my team, the cost of compute is far beyond the costs of the employees,” says Bryan
reached, and the AI agents were, in many circumstances, just as effective at closing a sale. “AI is revolutionizing the sales process,” says Aniketh Parmar, Centerfield’s chief technology officer. Stories like those have convinced plenty of leaders to commit, in some cases, millions of dollars to AI rollouts. Many firms also worried that competitors would beat them to the punch if they didn’t jump on the AI bandwagon.
Perhaps more important to many leaders was AI’s promise to cut most organizations’ biggest expense: people costs. Sure, some of those promises from AI-model developers seemed ridiculous—estimates suggested 40 percent of white-collar jobs would be eliminated—but non-tech leaders certainly believed that AI would let them trim their workforces, at least on the margins. Indeed, US business leaders believed that AI could, on average, lower their head counts by 1.2 percent by the end of the decade, according to an early 2026 survey conducted by the Federal Reserve Bank of Atlanta. That might not sound like much, but as a share of the country’s civilian workforce, it represents two million workers US firms wouldn’t need to hire, train, or pay. Leaders around the world subscribed to the same belief to varying degrees. Plenty of firms have used AI to justify cutting their workforces. In the US, about 15 percent of all layoffs over the last year were motivated by AI, and it was the reason for 40 percent of job reductions between March and June of this year.
More recently, when the World Wide Web was being built out in the 1990s, business leaders were promised that jumping online immediately and at at scale—and buying all of the necessary hardware and software—would lead to massive increases in sales and fat profit margins. In truth, many of those first-moving firms spent catastrophic amounts of cash (remember Pets.com or Webvan?) to attract customers. Then they collapsed, because things like warehousing and delivering goods to customers were hugely expensive (and the World Wide Web did nothing to lower the price).
Most of the current crop of business leaders were alive in the 1990s, and while they might not have been running firms back then, they should remember that the early hype of the web did not match the eventual reality. But AI’s siren song, in many cases, was enough for CEOs across industries to dive in headfirst. After all, AI could be the one technological breakthrough that would solve years of frustratingly slow revenue growth. And indeed, some of the early results have been downright magical for firms large and small. Waste Management, in its earnings call this summer, said that AI is generating $300 million in extra annual earnings by upgrading services and routing garbage trucks more effectively. Then there’s Centerfield, a marketing firm which helps connect customers to other organizations. It incorporated AI into its sales calls, then studied more than 260,000 customer conversations. Thanks to AI, the firm doubled the number of potential customers it
There’s always been a gap between what a technological
leap promises and what it actually delivers. During the 19th-century railroad expansion in the United States and Europe, promoters convinced small-town merchants that a rail line would turn their local burgs into bustling commercial centers where products could be sold to customers hundreds of miles away. But in many cases, the advent of the railroad backfired as trains simply flooded small towns with cheap, mass-produced goods from major industrial cities, forcing local merchants out of the business.
AI's
Siren Song
Still, savvy leaders have learned that AI can only take them so far. They need a human touch to excel, even if it costs them more money in the short term. Tarik Khribech found that a cheap monthly AI subscription lets him effectively run the website, marketing campaign, and several other things involved with the day-to-day operations of his company, AllBetter, which connects landlords and homeowners with contractors for repair work. After he turned the site over to AI, Khribech fired the various agencies helping him with web design and sales. But AI wasn’t improving the
Experts say that firms realize within six to 18 months of implementing a full AI rollout whether or not they got the AI/human balance right. If they got the balance wrong, it winds up being costly. In one case, Strout says, for every dollar a company saved when it laid off its 350 employees, it had to pay $1.27 to rehire them. Plus, there’s the opportunity cost firms faced not having humans in the organization to do the work when the work needed to be done.
The AI/
human balance
”
less people? We have no idea.”
To be sure, there are many leaders—in both the public and private sectors—who believe AI will still upend the global job market. The World Economic Forum estimates that AI and other automated systems will displace 92 million jobs by the end of the decade. But the group also says the technological shift will create 170 million new roles. “Do we need more people? Do we need less people? We have no idea,” says Paul Osterman, a professor emeritus at MIT and the author of Disposable Workers: The Transformation of Employment. At least for now, most employers aren’t
exactly racing to rehire. The $32 trillion US economy has only added, on average, 16,000 jobs a month over the last year. Outside of the COVID-19 pandemic, that’s the lowest rate since the Great Recession.
site’s appeal or services—both of which Khribech needed in order to attract more customers. It wasn’t long before he brought in different agencies—run by people—to help fine-tune AllBetter’s operations. “The AI model gives you the big picture,” Khribech says, “but for little details, you need the humans."
the human.
details, you need
For little
rediscovering something
It's the market
what AI can and
at scale, of
first experience,
We’re getting our
Do we need more people? Do we need
WATCH
PODCAST
How to Build an AI-Ready Workplace
Related Insights
READ FULL
ISSUE
Workforce 2026:
Growing Nowhere
Case Study: Korn Ferry’s Human + AI Journey