AI Now Covers the Equivalent of 12% of U.S. Work Hours, MIT Study Says
Work News | New Stardom
Trend Analysis
Photo by BoliviaInteligente
A new MIT report estimates that current Artificial Intelligence systems can already perform tasks equal to roughly 12 percent of U.S. work hours, a finding that places measurable boundaries around how far automation has actually progressed while signaling where substitution is becoming feasible.
The study, published this week by MIT’s Future of Work initiative, examines task-level exposure instead of job-level forecasts, mapping commercially available AI systems onto thousands of routine activities across the U.S. labor market. Researchers concluded that most tasks remain too complex, too context-dependent or too costly to automate end-to-end, although narrow functions in clerical, data-processing and customer-support roles can be carried out by existing models.
MIT’s team notes that the 12 percent figure reflects practical deployment, not theoretical capability, since many tasks that could be automated are not implemented due to integration costs, accuracy concerns or regulatory constraints. The report argues that automation pressure will continue to fall unevenly across sectors, as firms tend to adopt AI first in areas where the economic case is immediate and the required oversight is limited.
The researchers also found that large segments of healthcare, skilled trades, education, logistics, field work and public-facing services remain resistant to substitution because they involve physical dexterity, interpersonal judgment or compliance-heavy decision-making. Even in occupations with high task exposure, the report says AI is more likely to change workflows than replace full roles, since most jobs combine automatable and non-automatable components.
The study arrives amid rising concern about layoffs attributed to automation announcements, although the authors point out that productivity gains can take years to translate into net workforce reductions. They also caution that overstating capability can distort policy debates, as the gap between headline claims and documented performance remains wide.
MIT’s findings place the baseline lower than some industry projections that suggest imminent large-scale displacement, yet higher than assessments that classify AI as primarily assistive. The report frames the next phase of adoption as a management question rather than a technological one, since firms will need to redesign workflows and assign oversight before automation can scale.
Follow global work and job trends. Subscribe to New Stardom’s newsletter.
This work is funded directly by readers who value independent reporting.
If you find this useful, you can support New Stardom’s independent journalism.
Read Next
Have insights on work and the future of work? Submit an opinion piece to New Stardom. Love work and career books? Explore our fun workplace book collection.
New Stardom is an independent magazine covering the Future of Work, AI, and emerging job trends.
