The AI Job Apocalypse Has Not Arrived

The talking point arrived early. AI is coming for the jobs, the warning ran, and in months rather than years. Three years on from ChatGPT's release, the data behind that prediction is starting to look strangely empty.

The Yale Budget Lab has been tracking AI's impact on the US labor market since 2023. Its latest update, drawing on Current Population Survey data, concludes that "measures of exposure, automation, and augmentation show no sign of being related to changes in employment or unemployment." Even within the unemployed population, the team finds no clear concentration in AI-exposed roles. The picture, the researchers write, is one of "stability, not major disruption at an economy-wide level."

The Federal Reserve Bank of Atlanta tells a similar story. Across four surveys of US firms, more than 90 percent of companies report no meaningful AI-driven employment change over the past three years.

The investor class breaks ranks with the doom narrative

Even Andreessen Horowitz, hardly an AI skeptic, has now publicly broken with the apocalypse framing. In an essay titled "The AI Job Apocalypse Is a Complete Fantasy", general partner David George calls the narrative "unhelpful marketing, bad economics and worse history." The economic argument rests on what economists call the lump-of-labor fallacy, the assumption that an economy has a fixed quantity of work and that any task automated must come at someone's expense. History does not bear that out: cheaper inputs expand markets and shift human work toward higher-value activities.

That a Silicon Valley megafund is now arguing against the most aggressive AI hype is worth noting. Marc Andreessen and his colleagues spent two years insisting AI would transform everything. Now their pitch is that the workforce will be fine, which fits more comfortably with a portfolio strategy that needs healthy consumer demand to keep firing.

The actual squeeze is generational

None of this is to say the labor market is unscathed. There is one place the warnings have proven well-founded, and it is concentrated. Stanford researchers have found that workers aged 22 to 25 in the most AI-exposed occupations have seen a 16 percent relative decline in employment since 2022. Job postings aimed at recent graduates have fallen by 16 percent on at least one major platform, while applications per opening have surged by 26 percent. Jeffrey Sonnenfeld and his Yale colleagues described the dynamic in Fortune as "low hire, low fire": companies are not firing in waves. They are quietly not hiring at the entry level.

What that suggests is something narrower and more interesting than the original prediction. AI is not eliminating careers en masse. It is eating the on-ramps, the entry-level seats that historically gave young workers a foothold. The headline apocalypse is not arriving. A specific, generational squeeze is, and it deserves its own conversation rather than being folded into the louder, less defensible one.