A new MIT study estimates that artificial intelligence systems could replace 11.7% of the American workforce today—representing roughly $1.2 trillion in wages. The findings challenge assumptions that the highest disruption risks lie within the technology sector, instead pointing to routine administrative, financial and professional services functions as the most vulnerable.

Conducted in partnership with Oak Ridge National Laboratory, the research uses the Iceberg Index, a simulation platform that maps more than 32,000 skills across 923 occupations and models interactions among 151 million workers. The approach shifts the discussion from theoretical long-term forecasts to present-day economic viability for automation.

Researchers say conventional workforce metrics miss the early signals of disruption. “When AI automates quality control in automotive plants, consequences spread through logistics networks, supply chains and local service economies,” MIT researcher Ayusha Chopra noted. Traditional measures, she said, capture impacts only after they manifest in employment data, rather than when AI capabilities first overlap with human skills.

The study highlights how public attention on tech-sector layoffs obscures a broader reality. These visible disruptions represent just 2.2% of total wage exposure—around $211 billion. The remaining 97.8% is concentrated in roles such as HR, logistics, finance and office administration, which have received far less scrutiny.

Geographically, the findings challenge assumptions about which regions face the greatest exposure. Areas with large concentrations of back-office roles—particularly parts of the Rust Belt—register significantly higher Iceberg Index values than tech hubs. Workforces in Ohio, Michigan and Tennessee show risk levels driven largely by cognitive tasks supporting manufacturing operations.

Researchers emphasize that the Iceberg Index is not a prediction of when displacement will occur. Instead, it provides a skills-based model of current AI capabilities, giving policymakers tools to simulate scenarios and stress-test potential interventions before committing substantial resources.

Several states already are integrating the framework into planning efforts. Tennessee became the first to formally incorporate the index into its statewide AI Workforce Action Plan. Utah and North Carolina are developing similar strategies after supplying local labor data to validate the model’s accuracy.

While the study underscores that major workforce realignments may be ahead, it stops short of forecasting a collapse of entire professions. Instead, it identifies specific task categories—particularly routine administrative and coordination functions—most susceptible to near-term automation.

The release arrives amid rapid advances in AI, as tools evolve from chat interfaces to autonomous agents capable of executing multi-step workflows. The shift is already reshaping how corporations assess the economics of human versus machine labor, with implications for compliance, legal operations and organizational design across major sectors.