Artificial intelligence could fundamentally change the nature of work and transform the U.S. economy. So its potential to alter the nation’s admittedly troubled fiscal trajectory may seem niche.
Yet a growing number of economists have started to anticipate that widespread adoption of A.I. could structurally improve the imbalance between what America spends and what it collects in tax revenue in the coming years — as long as the technology does not put too many people out of work.
That is because A.I. could help make workers more productive, in turn growing the economy more rapidly. Faster economic growth would generate more tax revenue, helping to close the widening gap between what the government spends and the taxes it brings in.
Not that this would amount to a total fiscal utopia. Economists still expect the gross debt, now at about $40 trillion, to continue to grow, just less quickly than it might have without A.I. Several factors could impede or even reverse any budgetary progress, including the possibility that the gains from A.I. are lightly taxed. And Congress could always splurge on spending increases or tax cuts, canceling out any fiscal gains.
“The faster productivity growth from A.I. is not going to solve our budget imbalance,” said Douglas Elmendorf, a former director of the nonpartisan Congressional Budget Office. “It will help the problem, but it will not be enough to solve it.”
The consequences of A.I. for the budget will largely hinge on whether — and to what extent — the technology reshapes the labor market.
The bulk of America’s tax revenue comes from the money people make working. That is true not just because most of the nation’s income in general comes from work, rather than from investments in stocks or other business profits. But labor income is also taxed at a higher rate than capital income, a reflection of a long-held belief in Washington that lower taxes on capital helps encourage more investment and economic growth.
So the scenario in which A.I. makes workers more productive without making their jobs obsolete would be a positive outcome for the budget. Wages and tax receipts would go up. Companies and investors would also make more money and pay more taxes, albeit at lower rates.
It becomes more complicated if A.I. replaces at least some human labor, all while supercharging the profits that companies and investors make. That would mean more of the nation’s total income comes from low-tax capital rather than labor. Overall tax revenue could still increase, but by less than if the labor share of income stayed what it is now.
In an analysis of several different potential fiscal outcomes, researchers at the Yale Budget Lab found that the revenue bump from rapid A.I.-induced growth could be about half as large in 2030 if a greater share of income accrued to capital.
Trump Administration: Live Updates
Updated
- Trump honors a teenage lifeguard and takes aim at a CNN reporter.
- Prosecutors argue that a woman charged with vandalizing a Washington memorial is a flight risk.
- Trump again votes in Florida by mail-in ballot.
Then there is the possibility that A.I. takes every person’s job, creates an even more economically stratified society and upends American taxation entirely. (Not to mention the possibility that A.I. becomes adept at creating novel strategies for avoiding taxes). This arguably dystopian scenario, while not the base case for many economists, has started to guide ideas for addressing even a much less drastic loss of jobs because of A.I., with options ranging from spending more on support for people out of work to redesigning the tax system.
As Dario Amodei, the leader of Anthropic, put it in a January essay, the world’s billionaires should support the creation of “a more robust tax policy,” writing, “If they don’t support a good version, they’ll inevitably get a bad version designed by a mob.”
Chief among these possible tax changes would be increasing taxes on capital, which could involve raising the tax rate on capital gains, imposing new taxes on partnerships or cracking down on international corporate tax maneuvers. There are also some Democrats who want to raise taxes on A.I. companies or activities specifically, a potentially politically resonant option that some tax experts discount in favor of broader changes to how capital is taxed.
But some economists see the tax preference for capital as sacrosanct and would rather backfill any lost revenue with a value-added tax, a levy similar to a sales tax that is common in advanced economies. Many public-finance experts have long hoped the United States would start taxing consumption and see it as a safe recommendation no matter what the future holds.
“The good thing is that it is a fairly robust policy recommendation, in that even in the current regime many economists favor consumption taxation,” said Lee Lockwood, an economist at the University of Virginia.
Of course, taxes are just one side of the fiscal ledger. Even if Congress does not dedicate significant new resources to something like a universal basic income, A.I. could still increase how much money the federal government spends every year.
If A.I. helps scientists develop better medical care, for example, the average American life expectancy might increase. While longer human lives would be a good thing, it would also come at a cost to the budget. The government would have to spend more on Social Security and Medicare to keep delivering those benefits to retired people for longer.
A boom in A.I. investment could also increase demand from companies for financing, raising interest rates across the economy. That could raise one of the government’s other top costs: the interest it pays on the debt. So far this fiscal year, which ends after next month, the Treasury has paid $1.2 trillion in interest, more than it spends on the military and most other government programs.
How these different factors for the fiscal system actually turn out is highly uncertain, just as the consequences of A.I. for the economy — and humanity — are uncertain. A wild card is how the political system responds to the changes A.I. ushers forth, including any new surge in fiscal health.
Ben Harris, the director of economic studies at the Brookings Institution, remembers attending a conference around 2001, after a technology-generated jump in productivity in the 1990s had helped create fiscal surpluses. A major focus of the conference was what would happen to the financial system once the United States retired all of its debt.
That never happened. Congress responded to the surplus by repeatedly cutting taxes, including as recently as last year. The country went to war in Iraq and Afghanistan. The debt grew quite a lot.
“The bottom line is that those projections proved so incredibly rosy that I will likely always question forward-looking estimates of outgrowing our fiscal imprudence,” Mr. Harris said. “That some commentators and A.I. optimists were again projecting unprecedented growth rates reminded me of the optimism of the early 2000s.”
In other words, even A.I. might not be able to make a free lunch.
The post An A.I. Tax Boom Could Curtail America’s Debt. But Not Solve It. appeared first on New York Times.
