Laying Off While Fighting for AI Talent: The Contradiction That's Pushing Workers to the Breaking Point

Two Headlines, Same Week

Last month, I read two reports in the same week that, placed side by side, sent a chill down my spine.

The first: IBM, Salesforce, and SAP each announced a new round of layoffs, collectively affecting more than 20,000 employees. Every official statement contained the same word — "AI automation." The second: Data released by LinkedIn showed that over the past 12 months, job postings with titles containing "AI Engineer," "Machine Learning," or "Prompt Engineer" grew by more than 60% year-over-year, with some technical roles commanding salaries 40% higher than two years ago.

Both of these things are simultaneously true. That is precisely the problem.

"AI Layoffs" Are Not an Endpoint — They Are a Filtering Mechanism

Having observed the AI industry for the past several years, I am increasingly convinced of one thing: this wave of layoffs is not, at its core, about cost-cutting. It is a large-scale restructuring of talent.

What companies are eliminating are roles that trade time for output — content moderation, entry-level data annotation, basic customer service, mid-level report compilation. These positions share a common trait: clear inputs, predictable outputs, repeatable processes. Which happens to be exactly what current large language models do best.

Yet those same companies are desperately recruiting people who can "command AI": AI product managers, prompt engineers, AI systems architects, and RLHF annotation specialists responsible for evaluating model output quality — a role whose annual salary in Silicon Valley has already surpassed $100,000.

This is not a contradiction. It is a filter. The problem is that nobody told the people being filtered out where the exam hall is, or what's being tested.

The Middle Layer That Got Backstabbed

The hardest hit are mid-career professionals, roughly five to ten years into their working lives.

They are not entry-level, so they lack the flexibility of starting from scratch. They are not senior executives, so they cannot cushion the blow through decision-making authority and professional networks. They sit squarely in the sweet spot most vulnerable to AI displacement — specialized enough that companies once relied on them, yet not scarce enough that companies hesitate to let them go.

According to the World Economic Forum's 2024 report, an estimated 83 million jobs will disappear globally by 2027, while 69 million new ones will be created — a net loss of 14 million. But that figure obscures a far more brutal reality: the jobs disappearing and the jobs appearing are simply not going to the same people.

You cannot take an editor with eight years of SEO content experience and expect them to slot directly into an AI training engineer role. The cost of skill conversion is far higher than any "transformation budget" that appears in a corporate earnings report.

The Corporate Narrative and the Reality Gap

There is a talking point I have been tracking closely. I call it the "AI retraining myth."

Whenever a major company announces layoffs, the press release invariably includes a line like: "We will invest X hundred million dollars in employee reskilling." Amazon has said it. IBM has said it. AT&T pledged a $1 billion "retraining initiative" as far back as 2018.

But when you follow up on the data behind these programs, a few common problems emerge: completion rates are low (the average completion rate for online courses does not exceed 15%), course content lags the market by 12 to 18 months, and the actual percentage of participants who successfully transition into technical roles is rarely disclosed publicly.

Retraining is, more often than not, a public relations move designed to make layoffs look "responsible" — not a genuinely effective, systemic solution.

The Blast Radius of This Contradiction

In a previous article, I described this as a social time bomb — but that piece focused on the spread of systemic anxiety. Here I want to be more specific: the blast radius is wider than most people expect.

It is not just the individuals who have been laid off. It also encompasses those who "haven't been laid off yet, but live in uncertainty every day." According to Gallup's 2024 workplace survey, more than 51% of employed workers globally say they are currently in a state of "quiet watching" — not quitting, but not truly engaged either, because they are not sure whether their job will still exist next year.

This collective form of workplace disengagement, and the erosion it causes to organizational productivity, may be harder to quantify — and harder to reverse — than the layoffs themselves.

So, Is There a Solution to This Contradiction?

My assessment: in the short term, there is no structural fix.

But a few things are quietly shifting the landscape:

• The democratization of AI tools: A growing number of people without technical backgrounds are starting to accomplish tasks with AI tools that previously required engineers. The curve of skill barriers is dropping rapidly.
• A redefinition of "AI collaboration competency": Some forward-thinking companies have already stopped asking "Do you know Python?" and started asking "Can you effectively collaborate with AI and exercise critical judgment over its outputs?" This is a relatively more transferable skill.
• Rising regulatory pressure: The EU's AI Act has officially taken effect, with certain provisions requiring companies to fulfill disclosure obligations before using AI to replace human decision-making. This will not stop layoffs, but it may raise the legal costs enough to give companies pause.

Finally, a Conclusion That Is Not Entirely Comfortable

I have seen too many people read "the AI layoff wave" and "the explosion of AI job openings" as two separate stories. They are not.

They are two sides of the same story: a system that is actively phasing out old producers while reserving seats for new ones. The question was never whether AI is stealing jobs — it is that this upheaval is happening simultaneously, with no social safety net in place.

The real danger of this contradiction is not that it will explode. It is that it is already smoldering, and most people are still waiting for an official statement to tell them "this is just a transitional period."