A new study led by MIT labor economist David Autor examines the historical relationship between technological innovation and employment, offering critical insights for the current age of artificial intelligence. The research, published in the Annual Review of Economics, challenges the assumption that technology simply replaces jobs while creating new ones. Instead, it reveals a specific demographic trend: historically, new forms of work have disproportionately benefited young, educated college graduates working in urban settings. Using granular data from the U.S. Census Bureau spanning 1940 to 2023, the authors analyzed over half a century of labor market shifts. They found that in 1950, roughly 7 percent of employees held jobs in specialties that had emerged since 1930. More recently, between 2011 and 2023, approximately 18 percent of workers were employed in lines of work introduced since 1970. In both eras, workers under the age of 30 were the primary beneficiaries of these new roles. The study highlights that new work is intrinsically tied to scarce expertise. When a technology or new industry emerges, specialized knowledge is in short supply, commanding a wage premium. However, this advantage is temporary. As the expertise becomes common knowledge or is automated, the scarcity value erodes, and the work eventually becomes a standard task. For example, driving a car or using word-processing software was once a high-value skill but has since become a basic requirement. This dynamic suggests that AI, while creating new roles, may also accelerate the commoditization of skills, potentially reducing the long-term wage premium for certain specialties. Beyond the demographics, the research underscores the importance of demand-side innovation. The authors discovered that government-backed expansions, particularly during World War II, were massive drivers of new work. Counties with new factories during the 1940s saw a surge in new occupations, with 85 to 90 percent of the new work from that decade being technology-driven. This indicates that innovation is often a purposeful, cumulative activity fueled by strategic investment rather than random technological breakthroughs. Regarding the current debate on AI, Autor notes that it is too early to determine its full impact on the workforce. While there is widespread fear that AI will rapidly erode tasks, the study suggests that eroding tasks does not necessarily mean eroding jobs. The critical factor is how society and policymakers choose to implement AI. The authors point to the healthcare sector, where over half of U.S. spending is public, as a prime example of where government demand could steer AI toward socially beneficial outcomes. Instead of purely automating jobs, AI could be deployed to allow workers of varying expertise levels to perform different tasks, thereby boosting productivity and creating new roles. Ultimately, the study concludes that the future of work depends on intentional investment and policy. If society continues to demand and invest in new technological applications, new specializations will inevitably emerge. The challenge lies in ensuring that the benefits of these new roles are distributed broadly and that the rapid evolution of skills does not leave workers behind. As technology continues to advance, the lessons from the post-war era provide a roadmap for navigating the transition.