Overview
Bank of America economists, citing data from the International Labour Organization (ILO), estimate that nearly one in four jobs globally - roughly 838 million positions - are exposed to generative artificial intelligence. The bank's note, authored by economists led by Benson Wu, identifies younger workers, women and higher-educated employees as the groups most exposed to technology-driven disruption.
Key findings and international variation
The BofA analysis measures exposure to generative AI across income tiers. High-income countries, where non-routine cognitive work is more common, have the largest share of jobs exposed at 33.5%. By contrast, the exposure rate in low-income countries is estimated at 11%.
Alongside these exposure estimates, the economists caution that wealthier economies are positioned to capture the greatest productivity gains from AI adoption. They also warn that firms that lead the AI buildout are likely to take "a disproportionate share of those gains."
Context on technological disruption and labor markets
BofA's note places AI within the broader history of technological change. The economists write: "History provides many examples, from the Industrial Revolution to the internet age, where technological advances destroyed jobs that couldn’t be replaced. Yet, after initial disruption, the economy created new jobs that did not exist in the first place." They underscore that concerns about mass unemployment are widespread but note such scenarios are not fully consistent with economic theory and the evidence observed to date.
Evidence on worker outcomes from past technology waves
The note references a recent analysis from Goldman Sachs that examined the labor-market consequences of prior technological shifts for U.S. workers. Goldman’s researchers analyzed four decades of federal data and tracked more than 20,000 Americans born between the 1950s and the 1980s. They focused on workers displaced from jobs affected by technological change - for example, telephone operators and typists - and compared their outcomes with peers displaced from more stable fields.
The Goldman analysis found that workers displaced from technology-hit occupations experienced both short- and long-term economic costs. Those workers took, on average, about one month longer to find new employment and suffered real earnings losses of 3% after securing a subsequent job. Over the following decade, real earnings for displaced workers grew nearly 10 percentage points less than for workers who never lost a job, and 5 percentage points less than for those displaced from other industries.
Goldman researchers attribute much of this persistent gap to "occupational downgrading," a process in which the value of displaced workers' skills erodes and forces them into lower-paying roles.
Implications highlighted by BofA
The BofA economists emphasize that exposure to AI is not uniform. Younger workers, women and higher-educated workers appear more likely to hold roles vulnerable to generative AI. High-income countries show elevated exposure because of their larger share of non-routine cognitive occupations, even as they stand to benefit most from productivity improvements. At the same time, the potential for leading firms to concentrate gains raises distributional concerns about who benefits from AI-driven productivity growth.
Limitations
The note and the studies it cites focus on exposure and historical outcomes; they do not present definitive projections of employment levels or detailed paths for labor-market adjustment. Where evidence is limited, the reports highlight uncertainty rather than definitive forecasts.