ET
A consensus has formed that while artificial intelligence may create new and better jobs, its threat to current job holders requires massive new government training programs, unemployment assistance, income supplement programs and even a guaranteed minimum income. Missing from this rush to expand the government’s social safety net is any recognition that previous efforts to cushion the transition from jobs of the past to jobs of the future have done little to benefit those making the transition—and have raised the cost for society as a whole.
Societal gains from technological change come from what the economist Joseph Schumpeter called “the wave of creative destruction.” The lost jobs and investments rendered unprofitable by new technology free up labor and capital that can be redeployed to produce new and higher-valued goods and services. The more seamlessly the transition from the old to the new, the greater the gain from the new technology. “American exceptionalism,” our ability to generate and sustain higher living standards, has come in part from developing new technology and benefiting from being the first to implement it, and in part from our ability to move labor and capital dislocated by the wave of creative destruction efficiently into higher and better uses.
On average, every month since 2000 some 5.1 million American workers were separated from their jobs or were laid off and more than 5.2 million new jobs were created. In 2025, three times as many Americans changed jobs as did workers in the European Union. So inefficient is the Chinese economy in dealing with creative destruction that most industrial subsidies in China are used to sustain noncompetitive businesses. In short, the U.S. channels the wave of creative destruction through the economic system more efficiently than any other country in the world, and we are constantly enriched by it.
Government programs have provided a cushion to displaced workers, but they have also impeded the transitions. The 1962 Trade Adjustment Assistance program, which provided training, job-search and income support to workers harmed by foreign trade, has provided benefits to more than five million people. Numerous public and private studies have highlighted TAA’s failure by comparing the transition of TAA beneficiaries with workers who lost their jobs during the same period but didn’t receive TAA. Studies by the Government Accountability Office, the Labor Department and the U.S. International Trade Commission agree that TAA is insufficient in supporting dislocated workers to re-enter the labor market. It didn’t improve earnings. Benefits were used mostly as income support, and nonparticipants were re-employed faster than those who participated in TAA.
Another example is federal unemployment insurance, which was adopted in the 1935 Social Security Act and significantly expanded over the ensuing decades. A classic 1988 study by the economists Lawrence Katz and Bruce Meyer found that for every week of extra benefits, the covered worker was unemployed for as much as an extra day. The Congressional Budget Office found that “many workers find jobs in the weeks immediately before and after their benefits run out.” While unemployment benefits clearly are valued by people who lose their jobs, public and private studies generally conclude that on average the longer unemployment insurance is provided, the longer the worker will remain unemployed.
With growing calls for permanent new subsidies and a guaranteed minimum income for AI-displaced workers, we should heed the lessons of the 60-year-old War on Poverty, which President Lyndon B. Johnson declared as “not a struggle simply to support people” but “to allow them to develop and use their capacities.” Yet as the annual federal welfare spending surged to more than $70,000 per poverty family, labor-force participation among able-bodied persons in the lowest income quintile collapsed to 36%, from 68% in 1967.
A feel-good expansion of our existing programs to address AI transitions could idle tens of millions of workers, squander much of the economic benefit we hope to derive from AI, and foster a dangerous “bread and circuses” political system in which those who have chosen to remain outside the labor force demand an increasing share of the benefits created by those who have chosen to work.
Fortunately, a great blessing of AI technology is that it holds out the promise of facilitating the transition of workers from the jobs of the past to the jobs of the future. AI has been used in individual job searches and has been employed by the Harvard Business School to facilitate job placement for its graduates, but the technology hasn’t yet been applied by government to assist either the unemployed or current welfare recipients.
A national AI-employment system could provide individual assistance to the unemployed: assessing their aptitudes and interest and matching them with emerging jobs. It could provide individual training and track job openings in real time. If combined with a mandatory work requirement, AI could provide the most effective worker transition in history. It could revolutionize welfare by testing aptitude, providing individual training, and matching able-bodied welfare recipients with available jobs. An AI-assisted transition can assure that fewer people are left behind and that the societal benefits are expanded and broadly shared.
Mr. Gramm, a former chairman of the Senate Banking Committee, is a visiting scholar at the American Enterprise Institute. Mr. Solon is a senior fellow at the Hudson Institute.