AI was supposed to be killing jobs. In Spring, the labor market is opening up instead

URL SCAN: Fortune
FIRST LINE: Private employers added 122,000 jobs in May, ADP reported Wednesday, beating the 110,000 economists expected...

THE DISSECTION

This article is a triage dispatch from the ward of a patient in denial. The headline performs the ritual of "AI killing jobs" skepticism, but the subtext reads as: please don't look too closely at the vital signs beneath this headline number.

The piece assembles the current Jobs Week data—ADP hires, JOLTS openings, wage data—and uses it to dismiss AI displacement concerns. Expert Nela Richardson calls May "the strongest month of hiring since January of last year." Expert Skanda Amarnath offers two non-AI explanations for the firming: (1) reversal of the post-pandemic hiring freeze, and (2) immigration headwind normalization.

What the text is actually doing: Using a single-month cyclical reversal to adjudicate a structural thesis. This is the economic equivalent of citing a patient's good day to dismiss a terminal diagnosis.

THE CORE FALLACY

Category Error: Cyclical Stabilization ≠ Structural Viability

The article conflates two entirely different timescales and mechanisms:

• Cyclical stabilization: Hiring freeze unwinds, immigration shock normalizes, one-time pent-up demand releases.
• Structural obsolescence: The mass employment → wage → consumption circuit severs as AI achieves durable cost and performance superiority across cognitive work.

These are not competing explanations. One explains the next 12-18 months. The other explains the next 10-30 years. The DT does not predict uniform, simultaneous collapse. It predicts terminal decline with transition niches and death lag. Current data is entirely consistent with both a cyclical bounce and structural death already in progress.

The article makes the same error that observers made circa 2007 dismissing housing collapse concerns because employment was still strong. Strong current data does not refute structural trajectory.

HIDDEN ASSUMPTIONS

1. Aggregate headcount captures work reclassification. It does not. A laid-off paralegal and a retained paralegal doing half the work look identical in the unemployment count. The hollowing of job content is invisible to the headline.

2. One good month refutes a multi-decade structural thesis. This is the epistemic core failure. The DT operates on a 10-30 year structural horizon. A single-month reversal of a hiring freeze is noise on that timescale.

3. The "Information" sector wage suppression is noise. Amarnath dismisses the only "clean AI story" in the data by complaining about BLS categorization. The sector includes software publishing, data processing, and telecom. The workers who kept those jobs got the slowest wage growth in the economy at 4.0%—below the 4.4% average for job-stayers. This is wage depression from AI-enabled labor market pressure. It is not noise.

4. Job openings surging = firms preparing to grow. Amarnath waves off the JOLTS data as "not getting any worse." But the article itself reports that hires actually fell while openings jumped 730,000. High openings + falling hires is the textbook signature of structural displacement: firms advertise positions they cannot fill at current wage levels because the available human labor cannot perform the required work at the price they're willing to pay.

5. The frozen labor market pattern is benign. Low quits + low hires + flat wage growth = workers too afraid to leave and firms too uncertain to commit. This is not a healthy labor market. It is a market in cognitive and economic paralysis, entirely consistent with AI uncertainty suppressing both labor supply (workers fear replacement) and labor demand (firms unsure what human roles survive).

6. Small firm hiring is a good sign. ADP reports companies with fewer than 50 employees added 67,000 of May's 122,000 jobs. The article frames this as positive breadth. But small firm job growth is disproportionately gig work, contract work, and precarious employment—exactly the labor form most vulnerable to AI-mediated displacement. This is not a sign of healthy employment; it is a sign of employment atomization.

SOCIAL FUNCTION

Classification: Transition Management / Ideological Anesthetic

This article performs institutional reassurance for a middle-class readership anxious about AI displacement. Its function is to:

1. Neutralize AI anxiety before it reaches the level that demands policy response.
2. Manage the cognitive dissonance of simultaneous "AI revolution" headlines and "labor market strong" headlines.
3. Legitimize the institutional expert class (Richardson, Amarnath) as the appropriate interpreters of economic data, not the workers experiencing the hollowing.
4. Delay reckoning with the structural implications of AI capability growth.

The article is a hospice care admission form dressed as a wellness check.

THE VERDICT

The killing blow is not coming. It is already here. You are reading its early clinical signs through the noise of a hiring freeze unwind.

The DT does not require mass simultaneous unemployment to be correct. It requires durable, accelerating erosion of the mass employment → wage → consumption circuit. Every data point in this article is consistent with that erosion:

• Information sector wage depression (4.0% vs. 4.4% stayers): The sector most exposed to AI cognitive work is already experiencing wage suppression.
• High openings + falling hires: Firms advertising positions they cannot fill at viable wages. This is not strength. It is structural mismatch.
• Low quits, flat quits: Workers have stopped gambling on transitions—fear of AI replacement suppresses labor market dynamism from the supply side.
• Small firm atomization: The hiring is happening in the most precarious labor segments.

The article's experts correctly identify the cyclical drivers. They are wrong—willfully, institutionally, or through insufficient structural framework—that these drivers explain away the AI displacement concern.

The AI capex boom is not creating jobs. It is eliminating the need for them.

The lag between AI capability growth and measurable employment displacement is measured in years. We are in year two. The Information sector wage data is your leading indicator. The JOLTS openings/hires divergence is your early warning signal.

This article will age like a 2006 editorial arguing that subprime was contained.

WHAT THIS ARTICLE ACTUALLY REPORTS

BOTTOM LINE: The article is reading the patient's heart rate (still beating) as evidence the patient is not dying. The patient is dying. The heart rate is compensating. The attending physicians are explaining away the other vital signs.