How will AI change job markets?

Artificial intelligence will reshape labor by changing which tasks are done by people, which tasks are assisted by machines, and where work is located. Early quantitative forecasts and later empirical studies agree that AI is not a single, uniform force but a set of technologies that substitute for routine work, complement nonroutine cognitive tasks, and create entirely new roles in development, oversight, and application. Carl Benedikt Frey at the University of Oxford and Michael Osborne at the University of Oxford estimated high susceptibility to computerization for many occupations, highlighting the scale of potential change, while more recent modelling emphasizes both displacement and emergence of jobs.

Occupational shifts and skill demand

Routine manual and clerical tasks face the most direct risk of automation, but AI also complements skilled knowledge work, making some experts more productive while reducing the need for junior or repeatable tasks. David Autor at the Massachusetts Institute of Technology has documented how automation contributes to job polarization, expanding high- and low-wage employment while hollowing out middle-skill occupations. Daron Acemoglu at the Massachusetts Institute of Technology and Pascual Restrepo at Boston University provide empirical evidence that automation can reduce employment and wages in affected local labor markets, underscoring that technological gains do not automatically translate into broad-based employment growth. Large-scale analyses by James Manyika at McKinsey Global Institute project scenarios in which hundreds of millions of workers worldwide could see their tasks automated by 2030, stressing the importance of workforce transition policies. At the same time, the World Economic Forum led by Saadia Zahidi at the World Economic Forum finds that automation will also create new roles in data science, AI governance, and human-machine interaction, implying demand for reskilling and continuous learning.

Geographic, social and environmental consequences

AI’s impacts will be uneven across territories and communities. Urban centers with clusters of technology firms and educational institutions are likelier to capture new, high-value roles, while rural regions and countries with limited digital infrastructure risk slower adaptation and greater displacement. Gender and cultural patterns of labor may amplify disparities when care, domestic, and informal sectors are less visible to automation-related investment. Environmental considerations intersect with labor change: Emma Strubell at the University of Massachusetts Amherst has drawn attention to the significant energy and carbon costs of training large AI models, which can affect local environments and energy policy priorities where data centers concentrate.

Policy and institutional responses will determine whether AI’s productivity gains become broadly shared. Evidence across academic and industry sources emphasizes scalable upskilling, portable benefits, updated education systems, and stronger labor-market institutions to manage transitions. Without such measures, societies risk exacerbating inequality and geographic divergence; with them, AI can raise productivity, generate new career pathways, and free human labor for tasks emphasizing creativity, care, and oversight that machines cannot easily replicate.