The Jobs Apocalypse Got Postponed — And AI/ML Engineers Are the Reason Why

For two years, the dominant story about artificial intelligence and work has been one of dread: chatbots writing memos, agents closing tickets, whole categories of entry-level white-collar work quietly disappearing. On September 4, 2026, The Economist published a piece that complicates that story considerably. Its headline says it plainly: "The jobs apocalypse is postponed. An AI jobs boom is here." For anyone hiring — or trying to get hired — in AI and machine learning, it's worth unpacking why.

The headline numbers

The immediate trigger for the piece was the latest US jobs report: the economy added 162,000 jobs in August 2026, comfortably beating expectations, with unemployment holding at 4.1%. That alone would be a modest, unremarkable data point in most years. What made it newsworthy is who showed up in the numbers. Young workers — the cohort every AI-doom forecast singled out as most exposed to automation — held up "remarkably well," in the Economist's words, rather than being the first to lose their footing.

That finding lines up with independent research the article draws on. The Yale Budget Lab found no meaningful change in occupational mix or unemployment duration through March 2026 for workers in roles with high AI exposure. Brookings reached a similar conclusion. And a Vanguard analysis went a step further, finding that wage and job growth actually increased over the past two years in the occupations most exposed to AI, compared with less-exposed roles. If AI were simply hollowing out jobs,
this is not what the data would look like.

Where the jobs are actually coming from

The Economist's more interesting argument isn't that AI has been neutral for employment — it's that AI is a net job creator, and the mechanism matters. By the article's account, AI-linked layoffs have run to roughly 200,000 positions. Against that, AI-linked job creation has generated on the order of 1 million positions — call it 800,000 net new jobs, not lost ones.

Those jobs aren't showing up where the doom narrative predicted. They're concentrated in the buildout: data center construction, power and cooling infrastructure, electrical and HVAC work, utility-system construction — reporting elsewhere puts the infrastructure-linked job count alone at roughly 300,000 by 2026. But the white-collar side of the ledger is growing too, and this is the part that matters most for anyone reading this blog: software developers, data scientists, cybersecurity professionals, and — the category we track most closely — AI and ML engineers are all showing up as expanding lines on the hiring chart, not shrinking ones.

That tracks with what we've been seeing in the market piece by piece all year: labs and startups bidding aggressively for research and engineering talent, "forward-deployed engineer" emerging as a distinct hot title, and even a lab as dominant as Google DeepMind watching its hiring-to-departure ratio compress as rivals out-hire it for ML and applied-AI roles. The Economist's macro data is the receipt for a trend hiring managers have felt anecdotally for months: the AI buildout doesn't run without engineers to build it.

"A delay, not a cancellation"

None of this means the anxiety was baseless, and the Economist is careful not to claim total vindication. Routine service and administrative roles are still declining. The article — echoing a line from Bloomberg's Conor Sen, quoted in coverage of the same data — frames the moment as "a delay rather than a cancellation": a period of relative calm in the labor market, not proof the disruption isn't coming. Even OpenAI's own leadership has walked back the more apocalyptic framing this year, acknowledging in public remarks that AI adoption hadn't eliminated as many white-collar jobs as originally feared — without ruling out that it still could, on a longer horizon.

There's a practical thread running underneath the macro story too, echoed in Wall Street Journal reporting on the same trend: employers who froze entry-level hiring last year on the assumption that AI tools could cover the gap are now hiring again, having discovered that AI still needs humans to operate, evaluate, and take responsibility for what it produces. That's not an argument against AI adoption — it's an argument for pairing it with the engineers who can actually deploy it responsibly, which is precisely the skill set this hiring cycle is rewarding.

What this means if you're hiring — or hunting

For companies building AI products right now, the signal is straightforward: this is not a moment to under-hire on engineering. The infrastructure is being built, the tooling is maturing, and the constraint increasingly isn't compute — it's people who can turn that compute into working systems. For AI/ML engineers weighing whether the field is as volatile as the headlines from 2025 suggested, the Economist's data offers a genuinely reassuring counterpoint: exposure to AI, so far, has correlated with more demand and better pay, not less.

The apocalypse narrative made for a better headline than the boom does. But the boom is the one showing up in the hiring data.

Source: "The jobs apocalypse is postponed. An AI jobs boom is here," The Economist