New survey has some important information – including some chilling predictions
A sweeping survey of economists, tech insiders and forecasters finds broad agreement that artificial intelligence will reshape the labor market -but there is still deep uncertainty about how fast and how badly.
The economists surveyed were sober, measured professionals trained to resist hyperbole. Yet when asked what would happen to the American work force if artificial intelligence advances as rapidly as some in Silicon Valley predict, many of their answers were striking.
Roughly 10 million people could permanently leave the labor force by 2050. Wealth would concentrate at levels not seen since before the Second World War. And the occupations most at risk -general clerks, data entry workers, customer service representatives and assemblers -are precisely the roles that form the backbone of many large corporate head counts today.
Read next: Workers Are Using A.I. They Don't Trust. That's a Problem for the C-Suite
These are among the central findings of a major new study published this week by researchers from the Federal Reserve Bank of Chicago, Yale, the University of Toronto and the Forecasting Research Institute, among others. The paper, which surveyed nearly 200 economists, artificial intelligence professionals, policy researchers and highly accurate "superforecasters," offers the most rigorous attempt yet to quantify what A.I. might do to the American economy -and what human resources leaders should be preparing for right now.
The Baseline: More Disruption Than the Headlines Suggest
Even setting aside worst-case scenarios, the survey's findings are sobering for anyone who manages large work forces.
The median economist in the study expects the United States labor force participation rate -the share of adults who are working or actively looking for work -to fall from its current level of 62.6 percent to 61 percent by 2030 and 58.3 percent by 2050. That decline, driven partly by an aging population and partly by automation, represents millions of workers stepping back from the labor market entirely.
Read next: Employers deploy job protection measures amid widespread AI layoffs
The more striking figure emerges when economists are asked to assume that artificial intelligence advances rapidly -a scenario in which A.I. systems surpass human performance across most cognitive and physical tasks by 2030. Under those conditions, their median forecast for labor force participation by 2050 falls to 55 percent.
To put that in historical context: the United States has not recorded an annual labor force participation rate below 60 percent since the 1960s, before women entered the paid work force in large numbers.
"Roughly half of that decline -equivalent to around 10 million lost jobs -is attributable to A.I. rather than demographics and other non-A.I. trends," the researchers write.
What the Economists Actually Expect
The study is careful to separate what experts believe will happen from what they fear could happen, a distinction that matters enormously for workforce planning.
Unconditionally–meaning, given their all-things-considered best guess about how the world is likely to unfold–economists do not predict catastrophe. Their median forecast for annual G.D.P. growth is 2.5 percent through 2030 and 2050, modestly above government projections but far below the transformative acceleration some technology leaders have promised. The unemployment rate is expected to remain broadly stable, hovering around 5 percent even by midcentury.
Read next: The career ladder is disappearing. Are taxes making it worse?
What is striking, the researchers say, is the gap between those reassuring median forecasts and the enormous uncertainty that surrounds them. When asked to assume rapid A.I. progress, economists' projections for labor force participation spread dramatically. Their 10th percentile forecast for 2050 falls below 45 percent -a level that would imply an unprecedented structural withdrawal from paid work.
"Experts have far less confidence about what would happen if A.I. proves truly transformative," the paper states.
For human resources leaders, that uncertainty is itself the message. Planning only for the median outcome may be a significant mistake.
The Participation Cliff: How A.I. Could Reshape the American Work Force
U.S. labor force participation rate, historical and forecast to 2050, by A.I. progress scenario. Shaded band shows the range of economist uncertainty under the rapid scenario.
Which Jobs Are at Risk–and Which Are Not
The survey asked economists to rank 43 major occupation categories by their expected employment growth between 2025 and 2030. The results offer a remarkably clear picture of where work force vulnerability is concentrated.
At the bottom of the rankings -occupations where a majority of economists expected employment to decline -were general and keyboard clerks, customer service clerks, numerical and material recording clerks, other clerical support workers and assemblers. These are roles defined by routine, repetition and information processing: precisely the tasks that large language models have demonstrated the clearest ability to perform.
Read next: Employers warned against workforce cuts based on AI's potential
At the top, where economists expected the strongest employment growth, were personal care workers, health professionals, personal service workers and protective services workers -occupations defined by physical presence, human judgment and interpersonal connection.
The pattern is consistent across both the economists' baseline forecast and their rapid A.I. scenario, suggesting that the directional risk to specific occupations is not in serious dispute, even if the pace remains uncertain.
For H.R. leaders, the practical implication is direct: roles built primarily around processing, recording, scheduling and routing information face structural headwinds regardless of how quickly A.I. capabilities advance.
Winners and Losers: Which Jobs Economists Expect to Grow or Shrink
Occupations ranked by economists' expected employment change between 2025 and 2030 (unconditional scenario). Bars show relative ranking — the paper presents ordinal rankings rather than precise percentage figures for each occupation. The 50% midpoint divides those where a majority of economists expect growth (right) from those where a majority expect decline (left).
The Inequality Question
The survey also confronted a question that has begun to loom over boardrooms and compensation committees: if A.I. creates enormous economic gains, who captures them?
The answer from the economists is uncomfortable. Under their unconditional forecast, the share of national wealth held by the wealthiest 10 percent of households is expected to rise from 71.2 percent today to 75 percent by 2050. Under the rapid A.I. scenario, that figure rises to 80 percent -a level of wealth concentration not seen in the United States since the late 1930s.
The mechanism is not complicated. A.I. raises the returns to capital -to the ownership of data, compute, intellectual property and equity in technology companies -while compressing wages in occupations most exposed to automation. Workers whose skills complement A.I. tools may prosper. Those whose skills compete with them face pressure.
Labor's share of total economic output -the fraction that flows to workers as wages and salaries rather than to owners as profits -is forecast to fall from 55.5 percent today to 50 percent by 2050 under economists' baseline scenario, and to 45 percent if A.I. progress is rapid.
Some economists surveyed were more sanguine, arguing that competitive pressure among A.I. developers could turn artificial intelligence into a commodity whose gains flow broadly to users and consumers. Others warned that such an outcome was far from guaranteed and would require deliberate policy intervention to achieve.
What Policymakers–and Employers–Should Do
The survey asked respondents to assess the likely impact of six specific policy interventions, from retraining programs to universal basic income to a government "Manhattan Project" for A.I. development. For H.R. professionals, the findings offer a revealing window into expert thinking.
The policy with the broadest support among economists -71.8 percent in favor -was targeted retraining support: grants of up to $25,000 per year for workers displaced from high-automation industries, combined with career counseling and relocation assistance. Economists expected this to have a modest positive effect on both G.D.P. growth and labor force participation.
Universal basic income, by contrast, was opposed by 38 percent of economists and supported by only 37 percent, primarily because of concern that unconditional cash payments would reduce the incentive to work and further depress labor force participation.
The general public, however, saw things differently. A majority of ordinary Americans supported both a job guarantee program and universal basic income -policies that most economists opposed or viewed skeptically. That divergence suggests that if significant automation-driven displacement occurs, employers and policymakers may face political pressure that outpaces what economists consider optimal.
"If A.I. progress accelerates and labor market disruption becomes more visible, these underlying disagreements may become the central fault lines of policy debates," the researchers write.
A Nation Divided on the Response to A.I.: Economists vs. the Public
Share of respondents supporting implementation of each policy to address A.I.'s economic impacts. The gap between expert economists and the general public is widest on the most expansive proposals.
Preparing for Uncertainty
The study's most important finding for senior H.R. leaders may not be any single number. It is the shape of the uncertainty itself.
Economists' unconditional forecasts -the median outcomes they actually expect -are relatively clustered and not catastrophic. But the range of plausible outcomes under a rapid A.I. progress scenario is vast. Between the 10th and 90th percentile of economists' 2050 labor force participation forecasts under rapid A.I., the spread is more than 20 percentage points: the difference between a dramatic but manageable transition and a structural collapse of paid employment.
That is not a forecast of catastrophe. But it is a forecast of genuine uncertainty -the kind that demands scenario planning rather than point forecasts, and investment in work force adaptability rather than optimization for a single future that may not arrive.
The researchers who conducted the survey were explicit on this point. Even if the probability of rapid, transformative A.I. progress is only around 14 percent -as the median economist in the study estimates -the potential consequences are large enough to warrant serious preparation now.
"Policymakers cannot simply plan for the median outcome," the paper concludes. "They must contend with tail risks, including the potential for a deep contraction in labor force participation."
For the executives and H.R. leaders who will be managing those work forces, the same warning applies.
