Generative AI threatens 838 million jobs worldwide

The scale of the current shift is quantified by a stark figure: 838 million positions. According to data from the International Labour Organization cited by Yahoo 財經 and Bank of America, nearly one-quarter of all global jobs are now susceptible to the influence of generative artificial intelligence. This is not a uniform pressure; the risk is concentrated among specific demographics, most notably young workers, women, and those with higher levels of education.

This distribution creates a paradox. Those traditionally viewed as the most insulated—the highly educated—now find themselves on the front lines of technological exposure. This shift suggests that AI is impacting a wide array of unconventional cognitive work, affecting roles that have historically required advanced degrees and specialized training.

The exposure gap between high- and low-income economies

The risk of AI-driven displacement varies sharply based on national wealth and the nature of the local economy. In high-income nations, where unconventional cognitive work is more prevalent, the proportion of affected jobs reaches 33.5%. In contrast, the exposure rate in low-income countries is significantly lower, sitting at 11%.

This disparity is rooted in the structural composition of these workforces. Wealthier economies rely more heavily on the types of information processing and analytical roles that generative AI is designed to augment or replace. While this puts a larger percentage of the high-income workforce at risk, reports from Bank of America suggest that these wealthier economies are also positioned to capture significant productivity gains from the technology.

However, the distribution of those gains remains a point of contention. Bank of America suggests that the productivity boost may not be shared equally across the economy. There is a risk that the companies leading the construction and deployment of AI infrastructure will occupy a disproportionate share of these benefits, which could impact the overall distribution of wealth within the economy.

The lasting economic penalty of technological displacement

There is a prevailing economic theory that technological disruption is a temporary shock. From the Industrial Revolution to the rise of the internet, history shows that while technology destroys specific roles, the economy eventually generates new jobs that did not previously exist. Some economists and analysts suggest that the fear of mass unemployment is inconsistent with this historical evidence and existing economic theory.

“從全球來看,約四分之一的工作崗位面臨人工智能的影響。年輕勞工、女性及高學歷勞工受影響程度最大。” Benson Wu, Bank of America

Yet, the macro-level recovery often masks individual hardship. A Goldman Sachs analysis of forty years of federal data provides a more granular look at the human cost. By tracking more than 20,000 Americans born between the 1950s and 1980s, researchers found that workers displaced by technological change—such as typists and telephone operators—suffered far more than those who lost jobs in stable sectors.

The data reveals a persistent “income scar.” Workers in roles vulnerable to automation took one month longer to find new employment than their peers. Once they did find work, they experienced a 3% loss in real income. The long-term effects were even more severe: over the ten years following their unemployment, those displaced by technology saw real income growth that was nearly 10 percentage points lower than workers who were never unemployed, and 5 percentage points lower than those who lost jobs in other industries.

Defining the human boundary in professional standards

As AI begins to mimic high-level cognitive and creative output, professional institutions are beginning to establish hard boundaries to protect human labor. The film industry is currently a primary site for this struggle, as the distinction between human performance and digital synthesis becomes blurred.

The Academy of Motion Picture Arts and Sciences recently implemented new regulations to ensure that AI does not displace human recognition in the Oscars. Under these rules, AI-created actors and screenwriters are ineligible for nominations. The Academy stated that only real, living human performers, rather than AI avatars, qualify for acting awards. Similarly, scripts must be written by humans rather than chatbots to be eligible for writing categories.

These protections arrive as the industry tests the limits of digital resurrection. For example, a digital version of the late actor Val Kilmer, who passed away one year ago, was recently presented to cinema operators. The project, created with the support of Kilmer’s family using an image database, features a youthful digital twin in the trailer for the film As Deep as the Grave. While such technology offers new creative possibilities, the Academy’s ruling establishes specific criteria that distinguish human contributions from synthetic output.

This regulatory move reflects a broader effort to define the role of human skill and expertise in a professional landscape where the cost of synthetic alternatives is continuing to fall.

The central challenge of the AI transition is not the total number of jobs, but the quality and stability of the jobs that replace them. While the global economy may eventually balance the ledger, the Goldman Sachs data suggests that for the individual worker, the transition is rarely seamless. The risk is that displaced workers may face a long-term decline in earnings as they move from specialized careers into lower-paying alternatives.