The day the Fed said: growth might surge and jobs might vanish—and rate cuts won’t save you
The central bank’s language finally caught up to what many of you have been living. In a set of remarks that felt less like a speech and more like a dispatch from a moving frontier, Federal Reserve officials acknowledged that the AI reorganization of work is not a footnote to the business cycle; it may become the business cycle. The picture they drew is jarring: productivity climbs, output hums, and yet unemployment ticks up, not because demand collapsed, but because the way firms use labor has been rewired. That is not a problem you solve with a lower policy rate. It is the economy changing its operating system.
Governor Lisa Cook supplied the crucial sentence. Job displacement may come before job creation, she warned, and “our normal demand‑side monetary policy may not be able to ameliorate an AI‑caused unemployment spell without also increasing inflationary pressure.” In plain terms: if companies are shedding roles to reconfigure around AI, cutting rates is like pushing the gas pedal when the transmission is being replaced. You might rev the engine, but you don’t move the car. You just flood it.
Markets have been primed to see AI as a blanket disinflationary force—more code, fewer costs, lower prices, victory for doves. Kevin Warsh, the Fed chair nominee, has leaned into that thesis, arguing that productivity gains argue for easier policy. But the Fed’s conversation, as reported this weekend, is wider and more agnostic than that tidy story. Adam Posen called AI a “positive, real shock” with “very little disinflation,” and said those betting on near‑term price relief “have got it exactly wrong.” Positive real shocks raise capacity over time, but they also require investment, churn, and pricing power reshuffles in the middle. The bridge, not the destination, is the problem for a rate‑setting committee.
This is where the unemployment rate, a number that normally speaks in a single register, starts double‑booking its meanings. The Fed’s reference point for the “natural” rate sits around 4.2%. Above that, we have customarily read slack. But if AI pushes firms to restructure, the measured joblessness can rise without the usual collapse in demand, and therefore without the usual disinflation that would justify a policy rescue. Cook’s point turns the typical logic inside out: the same unemployment rate may no longer say what we thought it said about overheating versus slack. A dashboard light that once meant “slow down” might now mean “reprogram the engine.”
It would be academic if the reorganization were abstract. It is not. Block announced that it will cut roughly 40% of its workforce and credited the decision to a change in how it uses labor due to AI. That is not a factory installing robots; it’s a white‑collar platform, exactly the cohort that once assumed automation would nibble the edges first. When the Fed looks at that kind of announcement, it sees not just a company, but a pattern: skills mismatches, temporary idling of displaced workers, and productivity rising before labor is fully reallocated. Monetary policy can’t train anyone to prompt an agent, refactor a workflow, or validate a model’s output. Lower rates are a blunt instrument against a surgical problem.
If you want to understand why the Fed sounds unsettled, listen to Richmond’s Tom Barkin: “There are lots of forecasts… and the only thing you know for sure is those forecasts are going to be wrong.” That is not a throwaway line. It’s a recognition that the sign of AI’s near‑term effect on inflation is uncertain, and the lag structure is unknowable. Some adoption paths compress margins and tame prices; others ignite investment booms, bottlenecks for specialized talent, and pricing reshuffles as firms race to pay for capability before it pays for itself. The committee is used to chasing demand; it is less practiced at adjudicating a supply‑side rearchitecture in real time.
So what changes when the Fed starts treating AI job losses as structural rather than cyclical? First, the bar for rate cuts in response to a softer labor market gets higher if that softness arrives with solid productivity and output. Second, the burden of cushioning workers shifts conspicuously away from the Eccles Building toward tools that actually move people across occupations: training, credentialing, relocation assistance, safety nets that buy time without ossifying old roles, and industrial policy that accelerates productive diffusion rather than preserving incumbency. The message, between the lines, is: don’t wait for monetary easing to fix a reorganization problem.
This reframing also alters the political economy of patience. If AI‑linked churn pushes the unemployment rate up while the aggregate data still glow—unit output improving, corporate narratives bragging about efficiency—pressure will mount to “do something.” The Fed is signaling that its something may be restraint. That can look callous unless it is accompanied by visible, non‑monetary scaffolding for the transition. Absent that, we will be living with a paradox: a faster frontier for production paired with a rougher on‑ramp for workers, and a central bank that refuses to equate pain with stimulus if the pain comes from rewiring rather than recession.
The Phillips Curve with a new accent
At bottom, this is a story about translation. For decades, inflation and unemployment spoke to each other through a fairly legible grammar. AI introduces an accent. A rise in measured joblessness may not mean slack; it may mean reallocation. Higher productivity may not deliver immediate price relief; it may fund the capex and capability buildout that makes the relief possible later. The Fed is racing to build models that can parse those sentences—its research output on AI has visibly climbed since 2022, and January’s minutes carved out space for productivity and AI. But as Barkin conceded, even the best models will err in one direction or the other. The committee’s task is to choose which error is less damaging while the grammar is still changing.
If you’re a worker or a manager reading this, the practical translation is blunt. If unemployment rises because your firm is redesigning jobs around AI rather than cutting because demand evaporated, don’t expect rate cuts to be your parachute. Expect the conversation to pivot to whether you can be retrained, redeployed, or replaced. And expect the Fed to keep its eye on inflation risks born of investment and transition, not just on the grief inside the layoff statistics.
The last decade conditioned us to watch the Fed as the primary first‑responder to labor market pain. Yesterday’s debate suggests a new choreography. The central bank is preparing to keep its powder dry in the face of AI‑driven churn, even as it cheers the long‑run productivity boom. That’s not indifference. It’s a boundary. And it hands the real work of adaptation to the rest of us.
