The Massachusetts Institute of Technology on Wednesday released research estimating that AI systems are technically capable of performing work equal to 11.7% of the US labor market.
The study suggests that the exposed share represents approximately $1.2 trillion in wages across finance, healthcare administration and professional services.
The findings are based on the Iceberg Index, a labor-simulation model developed by MIT in collaboration with Oak Ridge National Laboratory (ORNL).
The model represents 151 million US workers as agents with more than 32,000 skills distributed across 3,000 counties, comparing those skills against the capabilities of existing AI tools.
“Basically, we are creating a digital twin for the US labor market,” research co-leader and ORNL director Prasanna Balaprakash told CNBC.
According to the report, the 11.7% figure reflects “technical replaceability,” defined as tasks that current AI systems could perform through a computer interface at a cost comparable to human labor.
The study does not project how quickly adoption might occur or how many workers could be displaced in practice, noting that technical capability is distinct from hiring decisions or implementation timelines.
The researchers found that exposure is concentrated in cognitive, desk-based work, such as finance and insurance, back-office banking functions, administrative and diagnostic tasks in healthcare, and professional and business services involving reporting, documentation and routine analysis. Jobs requiring physical presence or manual work, including many trades, hospitality roles and frontline care, were identified as less replaceable by currently available AI tools.
The Iceberg Index is designed to provide a forward-looking measure of where AI capabilities overlap with human skills before workplace changes occur. The researchers believe that the system may help policymakers and industry assess where retraining or investment could be needed.