AI Is Turning Workplaces Into Hopeless Gridlock Looks like AI is not the magical tool that CEOs make it out to be.

LinkedIn’s Blake Lawit, the chief global affairs and legal officer of the Microsoft-owned professional networking site, confirmed in an interview at the Semafor World Economy summit this week that the company’s data shows a decline in hiring of around 20% since 2022.

However, he pushed back at the idea that AI was to blame.

“At LinkedIn… we have an economic graph which is over a billion members. We’ve got companies, jobs, and skills. It’s really an amazing real-time view of what’s happening in the labor market. And we’ve looked — because everyone wants to know the answer to this question: Is AI impacting jobs right now? We’ve looked and, honestly, we haven’t seen it,” he said during his interview.

Instead, the executive suggested that the decline in hiring was more closely tied to a rise in interest rates.

“We have not seen the sort of impacts that you would expect to see in areas that everyone is talking about AI… like industries, whether or not it’s customer support, or administrative, or marketing — all these places that if we were seeing impacts [from] AI, that’s where it would be,” Lawit continued.

“Yes, hiring’s down, but not down more,” he added.

Lawit also noted that LinkedIn’s data didn’t indicate that the decline in hiring of college-aged young adults getting their first jobs was “down more,” either, when compared with people who were in the middle of or later in their careers.

Still, he didn’t rule out that things could change.

“Doesn’t mean it’s not going to happen in the future, but not yet.”

On that point, however, Lawit had a warning of sorts. Lawit noted that over the last several years, the skills that are needed to do the average job have changed 25%. With the rise of AI, LinkedIn expects that figure to be 70% by 2030.

“So, even if you’re not changing jobs, your job’s changing on you,” he said.

The AI industry’s "dirty little secret" just spilled into the open, and it isn’t pretty. 📉💻

A massive data breach at San Francisco startup Mercor is exposing the fragile, often exploitative underbelly of how the world’s most famous AI models—like those from OpenAI, Anthropic, and Meta—are actually built.

Here is the breakdown of the controversy shaking Silicon Valley:

1. The "Grim" Cottage Industry

Mercor has built a business model on the backs of a "disturbingly educated" but underemployed workforce. Desperate job-seekers are being hired to train AI models to perform the very tasks they can no longer find traditional work for.

The Conditions: Reports describe "crushingly long" shifts, inexperienced management, and contracts that vanish overnight without warning.
The Secrecy: Workers are often kept entirely in the dark about which tech giant’s AI they are actually training.

2. The Security Collapse

Late last month, Mercor revealed it had been hacked via an exploit in an open-source project. The fallout is significant:

Stolen Data: Leaked samples include Slack logs and videos of conversations between workers and AI systems.
The Risk: This doesn't just expose the contractors; it potentially leaks the "secret sauce" and proprietary training methods of companies like OpenAI and Anthropic.
The Corporate Panic: Meta has officially paused all work with Mercor while conducting its own investigation.

3. Legal Firestorms

The startup is now drowning in litigation:

Privacy Suits: Five new lawsuits accuse Mercor of leaking sensitive personal info, including Social Security numbers and home addresses.
Labor Disputes: Three separate class-action suits allege the company misclassifies workers as independent contractors to strip them of agency and benefits.
The "Switcheroo": Some contractors claim they were fired, only to be immediately offered the same work on a different project at a significantly lower hourly rate.

The Bottom Line

While the tech giants are primarily worried about losing their competitive edge through leaked training data, the human cost is becoming impossible to ignore. We are seeing a "fragile supply chain" built on the exploitation of the very experts AI aims to replace.

Is the "AI Boom" sustainable if it relies on a foundation of underpaid, overworked contractors?

CEOs have enthusiastically embraced AI as a silver bullet for office efficiency—often wielding it as a justification for sweeping layoffs. But a growing chorus of remaining employees reports a different reality: instead of lightening their load, AI is flooding their inboxes with error-prone, superficial content they must now fix. The result? More work, not less.

This phenomenon has a name: "workslop." Defined as AI-generated content that looks polished but lacks substance, workslop is creating hidden costs for companies racing to adopt the technology. According to The Guardian, a recent survey of 1,150 knowledge workers found that 40% had encountered work slop in their daily tasks—forcing them to spend an average of 3.4 hours per month correcting, rewriting, or discarding AI output. At scale, those minutes add up: for a company with 10,000 employees, that translates to an estimated $8.1 million in lost productivity annually.

The data aligns with broader research. One widely cited MIT study found that software developers actually became slower when using AI coding assistants. Another analysis revealed that 95% of companies deploying AI see no measurable revenue gain from its adoption—despite sky-high enthusiasm from executive suites.

The Human Cost: Stories from the Front Lines

Consider the experience of a copywriter at a Miami cybersecurity firm, who spoke to The Guardian on background. After his company laid off several colleagues and mandated AI use for content creation, he and his remaining teammates discovered a frustrating truth: while AI could generate seemingly professional drafts in seconds, those drafts required extensive human revision to be usable.

"Quality decreased significantly, time to produce a piece of content increased significantly and, most importantly, morale decreased," he said. "Everything got a whole lot worse once they rolled out AI."

The problem isn't confined to marketing departments. Philip Barrison, a sixth-year MD-PhD student at the University of Michigan Medical School, conducted a survey revealing that healthcare workers are also spending valuable time correcting flawed AI-generated patient communications—sometimes resulting in incorrect or confusing messages reaching vulnerable people.

The Perception Gap

These anecdotes point to a deeper disconnect: the chasm between executive expectations and worker reality. In a survey of 5,000 office employees, 40% reported that AI did not save them time. Yet among executives, 92% believe AI has made them more productive.

This dissonance matters. Employees doing the actual work understand that tasks requiring precision, nuance, and contextual judgment still demand human discernment—skills current AI tools cannot reliably replicate. That's why adoption remains uneven and why those closest to production work often hold mixed or skeptical views of AI's promise.

A Question Worth Asking

If frontline workers find that AI cannot reproduce their output at the same quality level, while CEOs—who rarely engage in the detailed execution of work—report productivity gains, a logical question emerges:

If AI can't replace the workers doing the work, but executives believe it makes them more efficient… could it be that the roles most vulnerable to automation aren't the ones being cut?

Some AI researchers are beginning to pose this question aloud. As evidence mounts that knowledge workers provide irreplaceable value—particularly in reviewing, refining, and applying judgment to AI output—it's becoming clear that the "lifeblood" employees of any organization can't be swapped out for bots without high cost.

The lesson for leadership is straightforward: before using AI as a rationale for headcount reduction, companies would do well to listen to the people actually using the tools. Efficiency isn't measured by how much content a bot can generate, but by how much value gets delivered—and that still requires skilled humans in the loop.