Total demand in the economy is determined by monetary policy, not technological change.
Total demand in the economy is determined by monetary policy, not technological change.
T wo widely read essays in recent weeks have warned that artificial intelligence will do more than eliminate jobs. It will, we are told, wreck the economy by destroying economic demand.
“The 2028 Global Intelligence Crisis,” published by Citrini Research, envisages AI replacing workers across sectors, slashing wage income, gutting consumer spending, and triggering a downward spiral of weak demand and rising unemployment. Another essay, Matt Shumer’s “Something Big Is Happening,” argues that AI’s advance could devastate white-collar employment and leave the broader economy reeling without the consumption it funds.
It’s an arresting narrative. It’s also wrong.
Artificial intelligence will likely cause significant sectoral disruptions. It may reorder labor markets, compressing some wages while elevating others. AI could accelerate productivity growth and intensify competition. But the claim that AI will cause a sustained shortfall in aggregate demand rests on a misunderstanding of how the economy works.
Aggregate demand, or total-dollar spending in the economy, is ultimately a monetary phenomenon. In the United States, aggregate demand is determined by the interaction of private spending decisions and the stance of monetary policy. The Federal Reserve, through its influence on the money supply and broader financial conditions, determines the path of total-dollar spending over time. Contracting demand is thus a failure of monetary policy, not a consequence of technological improvement.
The AI-doomer story goes like this: Firms replace workers with AI. Displaced workers lose income, so they cut back on consumption. Aggregate demand falls. Output and employment follow suit.
Yet this reasoning overlooks what happens next. When firms substitute AI for labor, companies’ labor costs fall, causing profits to rise in equal amount. Those profits accrue to shareholders, pension funds, and other capital owners. The money does not vanish into thin air.
Higher profits count as income, too. That income is spent on consumption, direct investment, or financial assets that finance other people’s investment. A dollar less in economy-wide wages does not mean a dollar less in total income. It simply means that income shifts from labor to capital. We can debate whether that redistribution is desirable. But the claim that it causes aggregate demand to collapse is mistaken.
Some argue that workers spend more of each dollar than capital owners, so shifting income toward profits would reduce consumption. But that only matters if the Federal Reserve allows total spending to fall. The Fed can meet higher demand for cash balances (or, alternatively, a lower velocity of money circulation) with standard open-market purchases. Under a monetary policy that stabilizes total spending, savings don’t leak out of the flow of expenditures; they get channeled into investment. Overall spending remains steady. The composition of demand may change, but the total does not.
The counterargument is that higher capital income will end up sitting in bank vaults, rather than financing new investment. But the point of the banking system is to channel capital to productive uses. It is highly unusual for financial resources to sit idle. Those who point to the 2008 Global Financial Crisis are making the argument for us: Lackluster investment was due to a disastrous Fed policy innovation — paying interest on excess reserves — not any inherent flaw in banking. New technologies like AI do not cause aggregate demand failures. Bad central-bank policy does.
More generally, AI-induced structural change will shift spending away from some labor-intensive services and toward data centers, semiconductors, energy infrastructure, and software platforms. There will be winners and losers across industries. Some occupations will shrink as new ones emerge. There may, of course, be a lag between old jobs disappearing and new jobs appearing. But that can’t cause total spending to plummet unless economy-wide liquidity — which the Fed controls — dries up. Don’t blame the creativity of entrepreneurs and computer scientists for the failures of central bankers.
The Fed already has the tools to stabilize demand in uncertain times: meet an increase in money demand with an increase in the money supply via open-market asset purchases. Such a response is demand-stabilization 101. AI may create radical new possibilities on the supply side of the economy, but the demand side can stay business as usual.
We admit that sudden drops in spending may cause unemployment to rise. But that has nothing to do with AI. History shows that broad demand slumps stem from policy mistakes, not new technology. The Great Depression was not caused by labor-saving machinery, but by a severe monetary contraction. The 2008 financial crisis and ensuing recession had a similar cause. Although the monetary base skyrocketed and the broader money supply grew at a historically normal rate, these developments were insufficient to meet a sharp spike in money demand. Aggregate demand plummeted as a result.
The AI-doomer argument also assumes that consumption is the sole engine of demand. While roughly 70 percent of spending is classified as consumption, that doesn’t mean it drives the economy. Investment matters just as much, since a dollar of consumption and a dollar of investment both contribute the same to aggregate demand. If AI reduces consumer spending but boosts investment, the effects offset each other. Historically, major technological advances — electricity, the internal combustion engine, the internet — triggered investment booms, not busts.
Certainly, AI will pose serious challenges. Distributional effects may be severe, and some workers will struggle to retrain or relocate. Policymakers may need to address education, mobility, and the tax treatment of capital and labor. These are real issues. But a permanent collapse in demand isn’t one of them.
As long as the Federal Reserve keeps aggregate demand stable, AI poses no greater threat than any other labor-saving technology. It will change what we produce, how we produce it, and who earns what — accelerating growth in some sectors while shrinking others. But it won’t create a permanent spending shortfall.
The Fed often lags economic changes by a quarter or two, as it waits for more data to confirm a spending slowdown. There’s no reason to expect the Fed to react either faster or slower to AI than any other economic change in modern history. AI may transform the economy, but it cannot make us too poor to afford its bounty.