What the Aggregate Data Actually Shows

Despite endless headlines about AI decimating white-collar work, unemployment in the occupations most exposed to AI is actually lower than in less-exposed jobs. That's according to Bureau of Labor Statistics data, analyzed and cited by former BLS chief Erika McEntarfer, who now works at the Stanford Institute for Economic Policy Research.

McEntarfer told MIT Technology Review bluntly: "All of the available evidence to date suggests that AI's impact on current labor market conditions is likely small right now."

There's also ZERO evidence that workers are fleeing AI-exposed jobs for manual labor alternatives. That's the economic signal economists watch most closely — and it isn't firing.

So if you've been sold an imminent jobs apocalypse, the numbers suggest otherwise.

What Is Happening — And It's Specific

The Stanford Digital Economy Lab released a working paper in November 2025 that deserves far more attention than it's getting.

Workers aged 22 to 25 in AI-exposed occupations saw a 16% relative decline in employment following the spread of generative AI. Their older colleagues in the same fields? No comparable drop. Workers in low-AI-exposure entry-level jobs? Also fine.

The hit is surgical. It's targeting junior tasks — the exact grunt work that used to be how young people learned, proved themselves, and climbed.

An Anthropic report from March 2026 pointed to the same pattern, according to MIT Technology Review.

The Federal Reserve Bank of New York piled on with its own Q4 2025 numbers: the unemployment rate for recent college graduates hit 5.6%, and underemployment — graduates stuck in jobs that don't require a degree — climbed to 42.5%, the highest since COVID.

The Career Ladder Problem Nobody's Talking About

Brookings dropped a report on April 2, 2026 that reframes the whole debate.

Stop obsessing over individual job titles. Start worrying about career pathways.

According to Brookings senior fellow Mark Muro and his team, there are over 15 million workers without four-year degrees in jobs highly exposed to AI. Of those, nearly 11 million are in what Brookings calls "Gateway" occupations — jobs that have historically been the on-ramps to higher-wage careers.

Almost half of the pathways between those Gateway jobs and higher-paying "Destination" jobs are now highly exposed to AI. The highest-risk corridors are administrative, clerical, and customer service roles, concentrated in the Northeast and Sun Belt.

This is happening now, quietly, to people who never went to Stanford and won't make the think-piece circuit.

AI Agents Are Already on the Job — Right Now

While researchers debate macro trends, agentic AI has already deployed into actual workplaces.

Wired reported that AI agents from a company called ProCollect are now making debt collection calls — autonomously. The bot, which introduces itself as "Eve," knows your name, knows how much you owe, calls to collect, and handles basic negotiations without a human on the line. According to analysis by the Kaplan Group, AI debt collectors will become a nearly $16 billion industry within the next decade.

Debt collection, ranked by CareerExplorer in the bottom 1% of all professions for job satisfaction, is being automated first. But what comes next on that list?

Meanwhile, Anthropic's Claude Code — described by Wired as "paradigm-busting" — released its Opus 4.5 version, which Anthropic claims scored higher than any human candidate ever on the company's own engineering hiring exam. This is a direct statement about what this technology can now do versus a human being applying for the same job.

Peter Steinberger's open-source tool OpenClaw, built to leverage Claude Code, racked up 366,000 stars on GitHub by early May 2026. Market adoption is real.

The School System Is Lying to Students

Yale's Jeffrey Sonnenfeld and his team at the Chief Executive Leadership Institute published a blunt commentary in May 2026 calling out universities for performing curriculum theater.

A core course developer at one of the world's top three business schools admitted to the Yale team: faculty are building AI courses that go obsolete fast, AND a growing number of students already know more about these tools than their professors.

From Yale's data: computer science majors now have more trouble finding jobs than humanities majors, per recent New York Federal Reserve Bank research.

Schools are selling AI fluency as the ticket to employment security. The data says that's not obviously true — and chasing the credential without understanding the actual landscape could leave a generation worse off.

On the Other End: Philosophers Are Getting Hired

Want a genuine plot twist in the AI economy? Wired reports that Google DeepMind has at least 10 in-house philosophers and Anthropic has at least 4. Demand for ethicists, logicians, and people who've read Kant is real — because agentic AI systems making autonomous decisions raise questions that engineers aren't trained to answer.

Small market. But real.

What This Means for Regular People

If you're mid-career, the broad data says you're probably fine — for now. If you're under 25 and in a white-collar field, the math has quietly shifted against you. And if you're a worker without a college degree trying to move up through administrative or customer service work, the Brookings data says the ladder you're counting on is getting shorter.

No politician in either party is treating this as the emergency it is. The Trump administration is busy classifying AI protesters as potential extremists — Wired obtained over 1,000 pages of DHS, FBI, and fusion center reports focused on what federal agencies are now calling "anti-tech violent extremism" — rather than addressing the workforce policy gap.

The problem isn't whether AI will eventually disrupt the labor market. It already has a specific target: the people with the fewest options and the least runway to adapt.