AI news
AI model reviews
AI tools
How-to guides
Our staffAboutContact

Fed chair Warsh outlines three AI scenarios, including job losses and a stock slump

Fed Chair Kevin Warsh says Board staff are modelling three paths for "super intelligence", from a shared boom to higher unemployment to an investment bust. He calls them "not forecasts".

a large building with columns and a flag on the corner
Photo: Joshua Woroniecki on Unsplash

Fed Chair Kevin Warsh said on Monday that Federal Reserve Board staff have started modelling how "super intelligence" could change the US economy over the next several years. The scenarios include one where unemployment rises even as the economy grows, and another where an AI investment bust pulls stocks down.

He was careful about the framing: these are "illustrative scenarios, not forecasts."

The three paths Warsh described

Warsh was opening the Productivity and Jobs Task Force Conference, hosted by the Board and the San Francisco Fed. He calls the technology "super intelligence", or SI. Decrypt reports that's now official Fed terminology after an October change that followed a Trump administration executive order; it isn't a claim that today's systems have passed human intelligence.

In the best case, SI lifts productivity and the gains are shared across workers and the owners of capital. Real GDP growth stays persistently above its long-run average, and the US stays close to full employment.

The middle case is messier. An SI-driven expansion could displace jobs rather than complement them, and GDP growth would still beat its long-term average. But he said widespread automation, combined with labour market frictions, could push the jobless rate above levels consistent with full employment.

In the third, SI disappoints. After heavy capital investment and stock market gains, expected returns turn down and equity valuations fall, leading to a slowdown and a more notable rise in unemployment. In time, he said, the economy recovers, but without the productivity gains.

What the Fed is and isn't claiming

Warsh said plainly that his colleagues aren't there as futurists and aren't making predictions. He added that many other outcomes could be drawn, probably including the one that proves most accurate. The staff work uses stylized, model-based frameworks, which means simplified economic models rather than detailed forecasts.

He also described the Fed's role as "neither advocacy nor opposition" toward the technology. The big open questions, as he framed them, are magnitude, scope and timing: whether AI raises total factor productivity, when the surge starts and when it tapers. He flagged critical infrastructure too, since SI puts power in the hands of "all, including bad actors."

The speech didn't propose a rate change. Decrypt notes the Fed's next policy meeting is scheduled for December 8 and 9.

The money behind the boom

The third scenario isn't abstract, and a second Fed official spoke to it the same day. Governor Michael Barr, speaking in Milan, pointed to the Fed's latest Financial Stability Report, released on Friday, which focuses on the AI investment boom.

Going by Barr's account of the report, on-balance-sheet debt at the seven largest AI firms is expected to pass $800 billion before year's end. Their off-balance-sheet obligations, such as long-term compute leases, energy contracts and residual value guarantees, have a potential maximum nominal value above $3 trillion. Barr noted that firms usually present those figures before discounting, so they don't compare neatly with ordinary debt.

He wasn't alarmist. Given the still-limited debt backing the boom, he said he's less concerned about direct losses, though he warned that high hedge fund leverage could turn a sharp reversal into forced selling. He also said policymakers don't target market levels.

The stakes for people who use AI

If you use AI tools at work, none of this changes your subscription or your tools this week. What it shows is how the people who set US interest rates are thinking: a world where AI boosts output but not everyone's paycheck is on their planning list, and so is a world where the spending doesn't pay off.

I think the middle scenario deserves more attention than it gets. A boom that leaves workers behind sounds less dramatic than a crash, but it's probably the one that touches more households, and it can happen while headline GDP looks great.

Honestly, I'd treat the exercise as a sign of how uncertain the Fed is, not as a warning of what's coming. Warsh himself said he expects "much more clarity" over time, and I'd guess the staff scenarios will keep changing as the data arrives. The December meeting is the next public moment to watch, though nothing in Monday's remarks suggests AI is about to drive a rate decision.

Keep reading