The Massachusetts Institute of Technology released a study on Wednesday that found artificial intelligence could already replace 11.7% of the U.S. labor market, or $1.2 trillion in wages across financial, medical and professional services.
The study was conducted using a labor simulation tool called the Iceberg Index created by MIT and Oak Ridge National Laboratory. The index simulates how 151 million U.S. workers interact across the country and how they are affected by AI and corresponding policies.
The Iceberg Index, released earlier this year, offers a positive outlook on how AI has the potential to reshape the labor market, not just in coastal tech hubs, but across every state in the country. As lawmakers prepare to invest billions in reskilling and training, the index provides a detailed map of where disruptions are occurring, down to the zip code.
“Essentially, we are building a digital twin for the U.S. labor market,” said ORNL Director and study co-lead Prasanna Varaprakash. ORNL is a Department of Energy research center in East Tennessee that is home to the Frontier Supercomputer that powers many large-scale modeling efforts.
Varaprakash said the index conducts population-level experiments that reveal how AI reshapes tasks, skills, and labor flows long before the changes are visible in the real economy.
The index treats 151 million workers as individual agents, each tagged with skills, tasks, occupations, and locations. It maps more than 32,000 skills across 923 occupations in 3,000 counties and measures where current AI systems can already perform those skills.
The researchers found that the visible tip of the iceberg – layoffs and role changes in technology, computing and information technology – accounted for only 2.2% of total wage exposure, or about $211 billion. There is a total of $1.2 trillion in wages behind the scenes, including day-to-day jobs such as human resources, logistics, finance, and administration. These are areas that are sometimes overlooked in automated forecasting.
The index does not accurately predict when or where jobs will be lost, the researchers said. Instead, it aims to provide a skills-focused snapshot of what today’s AI systems are already capable of doing, and provide a structured way for policymakers to consider what-if scenarios before they actually commit funding or legislation.
Researchers partnered with state governments to run proactive simulations. Tennessee, North Carolina, and Utah helped validate models using their own labor data and began using the platform to build policy scenarios.
Tennessee moved first, citing the Iceberg Index in its official AI Workforce Action Plan released this month. Utah state leaders are preparing to release a similar report based on Iceberg’s model.
North Carolina Sen. DeAndrea Salvador, who has worked closely with MIT on the project, said what attracted her to the study was how it surfaced effects that were missed by traditional tools. She added that one of the most useful features is the ability to drill down to local details.
“It’s county-specific data that matches within a given census block the skills that are currently being performed and the potential for those skills to be automated or enhanced. And what that means in terms of employment as well as changes in the state’s GDP in that area,” she said.
Salvador said this type of simulation work is especially valuable as states are setting up overlapping AI task forces and working groups.
The Iceberg Index also challenges common assumptions about AI risk, namely that it is limited to the technology’s role in coastal locations. The index’s simulations show that occupations across all 50 states are exposed, including inland and rural areas that are often left out of AI discussions.
To address this gap, Iceberg’s team built an interactive simulation environment that allows states to experiment with a variety of policy measures, from changing labor costs and tweaking training programs to examining how changes in technology adoption affect local jobs and gross domestic product.
“Project Iceberg will enable policymakers and business leaders to identify infection risk hotspots, prioritize training and infrastructure investments, and test interventions before committing billions of dollars,” the report said.
Varaprakash, who also serves on the Tennessee Artificial Intelligence Advisory Council, shared state-specific findings with the governor’s team and the state’s AI director. He said many of Tennessee’s core sectors, such as health care, nuclear power, manufacturing and transportation, still rely heavily on manual labor and are somewhat insulated from pure digital automation. The question, he says, is how to leverage new technologies such as robotics and AI assistants to strengthen these industries rather than hollow them out.
For now, the team is positioning Iceberg not as a finished product, but as a sandbox that states can use to prepare for the impact of AI on their workforces.
“The idea is to really get involved and start trying out different scenarios,” Salvador said.
WATCH: Amazon targets middle managers with mass layoffs, memo hints at further job cuts as AI dilutes Big Tech
