Vanguard economist: AI job losses still in early phase, like ATMs in 1980s

Vanguard's chief economist uses the history of banking technology to frame a debate over AI and employment that has grown heated since ChatGPT's arrival in late 2022. The parallel begins in the 1980s, when automated teller machines became widespread. Many observers predicted bank tellers would disappear—but this proved only partially true. While the number of tellers needed at individual branches did decline as ATMs automated routine transaction work, the total employment of U.S. bank tellers remained broadly stable from 1980 through 2010. The reason: lower operating costs made it economical for banks to open additional branches, offsetting the efficiency gains from automation. More importantly, banks used these savings to expand into adjacent roles that required higher skills—loan officers, credit analysts, personal bankers, fraud specialists, and risk managers. Bank branches shifted from transaction hubs to relationship-management centers. The economist notes this created "new demand for more occupations" even as the ATM changed the nature of individual teller jobs. The real disruption, however, came later. Beginning around 2010, mobile banking changed the equation fundamentally. Unlike ATMs, which automated a single task within the branch, mobile banking largely automated the entire trip to the bank. The Electronic Signatures in Global and National Commerce Act of 2000 had already given digital signatures legal standing equal to ink signatures, enabling fully digital banking and accelerating the move away from in-person transactions. The impact was stark: by 2025, only 9% of bank customers reported branches as their primary banking channel, down from 36% in 2007. Bank teller employment fell accordingly. The economist concludes that "isolated task automation rarely results in large-scale job losses, except in occupations built around a very narrow set of activities," citing the obsolescence of switchboard operators as an exception. Instead, "meaningful disruption occurs when technologies are combined with new workflows, business models, and institutional changes that fundamentally alter how work is organized." Applied to AI, the economist argues that if it becomes a general-purpose technology like electricity or the personal computer, it will enable entirely new products, services, and industries—as mobile banking did, creating cybersecurity analysts, digital product managers, payment-platform engineers, and data-platform operators. The data since ChatGPT's launch supports this view. Nearly four years later, "occupations with the greatest exposure to AI have not experienced widespread employment declines." Employment growth in highly exposed occupations has generally kept pace with or exceeded that of less exposed ones. Layoff rates remain low, and although hiring has slowed, the slowdown has been broad-based rather than concentrated in AI-intensive fields. The economist thus suggests we remain "closer to the ATM phase than the mobile banking phase"—suggesting more significant disruption may require something closer to the deeper reconfiguration mobile banking triggered: a shift in business processes, organizational structures, and customer interactions that reshape rather than eliminate workers' roles.