AI Reality Check: Is the Economy About to Collapse?

Lately the headlines say AI is about to wreck the economy—mass layoffs, industries folding, white‑collar workers learning trades—and one World War Z–style memo even got blamed for a wobble in the S and P 500. How seriously should we take that story line? I am Cal Newport, and you’re listening to AI Reality Check.

AI coverage moves in waves; we’ve cycled from consciousness myths to superintelligence and now to mass job loss. An Atlantic piece pivots from BLS history to asteroid metaphors and claims that software learns as it goes, which misreads how today’s static language models actually work.

It props up the fear with headline layoffs and CEO pronouncements. Those cuts at places like Meta and Amazon mostly trace to pandemic overhire and failed bets like the metaverse, and AI bosses have every incentive to talk up world‑shaking power while their own revenue math is still upside down.

A Times op‑ed spotlights a grad struggling to land a desk job and links a cooler white‑collar market to generative AI. Economists point instead to pandemic‑era overstaffing followed by hiring pauses, higher rates that dampen expansion, and global uncertainty that makes leaders wait and see.

The op‑ed then urges us to take AI builders at their word about sweeping job losses. We should not, because hype makes their tools sound inevitable while masking unanswered questions about sustainable business models.

Then came a 2028 crisis “dispatch” that reads like financial fan fiction, pegging today’s layoffs to a future collapse after a short productivity boom. It spreads so well because it anchors fear to real events and tells the tale as if we’re already living the aftermath.

I felt that gut punch too, and I’m allergic to hype. But the people paid to be right about macro outcomes quickly pushed back.

A Deutsche Bank strategist said the piece leans on story over evidence, and a Fed governor waved off rapid white‑collar displacement. A Citadel Securities analyst basically quipped that if we can’t reliably forecast payrolls two months out, a Substack scenario will not map the next four years of labor destruction.

He noted no meaningful surge in day‑to‑day AI use at work in real‑time data. Diffusion follows S curves—slow, then faster, then leveling—not an endless vertical line.

Compute costs also set hard boundaries; as adoption rises, capacity tightens and prices climb, which often makes humans cheaper for many tasks. You can already see the economics in coding tools that are being sold below cost to grab share.

Labor indicators show little AI-driven disruption so far and some forward‑looking pieces have improved. The sensible base case is modest: AI likely offsets headwinds from aging, climate, and deglobalization more than it erases work.

Markets reflect that view; they are skeptical of frontier AI firms’ profit math, make only measured sector bets, and react far more to oil spikes than to apocalypse threads.

Could AI still cause painful disruption in places? Yes, but doom writing is counterproductive and lets executives dodge accountability. When a tech boss blames sweeping AI forces for cuts that really stem from bad crypto bets, reporters should press the business facts, not crown him a prophet.

Treat AI like a normal technology, track real impacts, and use normal policy and management tools to shape outcomes. That’s it for today—likely back next Thursday if there’s something worth unpacking. Take AI seriously, but not everything written about it.