Will AI Replace Jobs? What I Tell Leaders After Watching the Great Reversal of 2026

Let me give you the answer I give every founder and CEO who asks me this over coffee: AI is replacing tasks far faster than it is replacing jobs, and the leaders who confused the two are the ones quietly rehiring right now. I have spent over a decade building AI systems for companies across 20-plus countries, and 2026 has been the year my inbox filled with a specific kind of message. It usually starts with "we cut the team a year ago, and now we have a problem."

If you are reading this to take something useful away, here it is up front: do not make a headcount decision from an AI demo. I have watched that single mistake cost companies more than the AI ever saved them. Let me walk you through what I have seen, and what I would do in your seat.

What I mean by "the great reversal"

Here is what is happening on the ground. Through 2026, company after company is walking back the AI-driven layoffs they made so confidently in 2024 and 2025. They automated a function, cut the people, announced the efficiency win, and then hit reality.

I am not speaking only from my own client conversations, though I have had plenty. The wider data matches what I see. Two in three employers who made AI-driven layoffs are already rehiring, often within months, and around 55% of them now say they regret the cuts, according to 2026 workforce reporting. Forrester's Future of Work analysis expects roughly half of all AI-attributed layoffs to be reversed, frequently rehiring at lower wages. Hold onto that last point, because when I sit with a leadership team, it is the part they least want to hear.

When a CEO tells me AI let them cut a whole team, my first question is always the same: which tasks did AI actually take, and who was doing everything else? The pause after that question tells me whether they are about to join the reversal.

Why I watched smart leaders get this wrong

I want to be fair to the people who over-cut, because many of them are sharp operators. They got it wrong for an understandable reason: a job is a bundle of tasks, and AI is genuinely dazzling at some of those tasks. You see AI draft the email, summarize the report, answer the routine question, and your brain makes a leap from "AI did that" to "AI can do that job." I have felt the pull of that leap myself.

But here is what I have learned from actually shipping these systems. Take a support agent. They answer questions, yes, but they also handle the strange edge case, defuse the furious customer, spot the pattern that signals a bigger problem, and know when to escalate. AI can take the first, most repetitive slice. It cannot reliably take the rest. When a leader cuts the whole role because AI absorbed 40% of it, they find out very quickly who was quietly doing the other 60%.

This is the same reason most AI projects disappoint, not just staffing ones. Only about 13% of companies report a positive EBITDA impact from AI so far, even though nearly everyone agrees it lifts productivity. I say this to clients constantly: productivity at the task level does not automatically become profit or fewer people at the business level. The translation is where leaders stumble, and it is where I spend a lot of my time.

What AI is actually taking, in my experience

Let me be precise, because vague answers help no one. In the work I see, AI is replacing tasks, not people, and that distinction should shape every decision you make. The tasks most exposed are the repetitive, rules-based, high-volume ones: routine drafting, summarizing, first-line support, data entry, basic analysis. If a role is almost entirely those, it is genuinely at risk.

Here is how I describe what is really happening when leaders ask me to cut through the noise:

• Tasks are being automated, especially the predictable ones.
• Roles are being reshaped, as people move toward judgment, relationships, and exceptions.
• A few jobs are genuinely disappearing, where the role was almost all automatable tasks.
• New roles are appearing, to build, supervise, and correct the AI itself.

The people I watch thrive are not the ones hiding from AI. They are the ones who hand it the boring slice of their job and pour the freed time into the parts that need a human. I tell my own teams the same thing. That is reshaping, and it is the opposite of replacement.

The numbers I keep in front of me

I do not run my company on vibes, and I would not ask you to either. So here are the figures I keep in front of me when I think about AI and jobs, because they tell a more honest story than the headlines.

The World Economic Forum's Future of Jobs Report 2025 projects that by 2030, technology and other shifts will create about 170 million new roles while displacing roughly 92 million, a net gain of around 78 million jobs. That is real disruption, roughly 22% of all jobs churned, but it is churn and creation, not a one-way collapse. When I show a nervous leadership team those numbers, the conversation changes from "how many people can we cut?" to "how do we move people toward the roles that are growing?"

A few more that shape how I think:

• 39% of core job skills are expected to change by 2030 (WEF), down from 44% in 2023. The skills shift is real, but it is not accelerating out of control.
• 85% of employers say upskilling is their primary workforce strategy for 2025 to 2030 (WEF). The smart money is retraining people, not just removing them.
• 63% of employers name the skills gap as their biggest barrier to transformation (WEF). The bottleneck is talent that can work with AI, not a surplus of workers to shed.
• Only about 13% of companies report a positive EBITDA impact from AI so far, despite near-universal productivity claims. Value is lagging the hype.

Put those together and the picture is clear to me: the constraint most companies face is not too many people, it is too few people who can turn AI into results. That is the opposite of the assumption behind the 2024 and 2025 layoffs.

The detail I make every leader sit with

If I could get one fact into every boardroom, it would be this: the rehired roles are often coming back at lower wages, and that is not a happy ending. It tells me that even when a company admits it cut too deep, the damage lingers. They churned their people, lost institutional knowledge, spent trust they cannot easily rebuild, and paid twice, once to cut and once to rebuild.

So when a leader tells me AI is their chance to slash payroll, I push back. I have seen where that road ends. The companies rehiring at a discount did not win a cost battle; they created disruption they are still paying for. And for the workers, the lesson is just as clear: the ones who stayed valuable were doing work AI could not easily absorb. If you are trying to gain something from this piece, gain this: treat "automate and cut" as the expensive playbook it has proven to be.

What I would do if I were in your seat

Here is the advice I actually give, stripped of hype. Do not use AI as a blunt instrument to shrink the team. Use it to expand what your team can do, and only then ask whether you need fewer people. In almost every case I have seen, that reframing, from replacement to leverage, is what separates the companies capturing real value from the ones drafting their reversal announcement.

Concretely, this is the discipline I would hold you to. Automate tasks, not roles, and watch closely what your freed-up people do with the time. Redesign roles deliberately around the higher-value work AI cannot touch, instead of deleting the role and hoping. Measure the actual business outcome, not the theoretical headcount saving, because I promise you the two will not match. And treat any AI-driven staffing cut as reversible until it has genuinely proven itself, because the data says a coin flip's worth of them get reversed anyway.

This is the same discipline my teams use when we build AI: start from the real problem, keep humans in the loop where judgment lives, and measure honestly rather than trusting the demo. The great reversal of 2026 is, to me, just a very large and very public lesson in what happens when that discipline gets skipped.

What I want you to take away

So, will AI replace jobs? In my experience it is replacing tasks at scale, reshaping most roles, eliminating a few, and creating others, but the wholesale replacement that the 2024 and 2025 layoffs assumed simply has not shown up. That is why so many of those cuts are being undone. The reversal is the market correcting an overconfident bet, and you do not have to make that bet yourself.

If you take one thing from me, take this: automate the task, keep the judgment, and prove the value before you cut a single role. The leaders who look wise in 2027 will not be the ones who deleted headcount fastest. They will be the ones who turned AI into leverage for their people and measured the result honestly. I have staked my own company on that belief, and everything I have watched this year has only made me more sure of it.