A new study released Wednesday by the Massachusetts Institute of Technology reveals that current artificial intelligence systems are already capable of performing work equivalent to 11.7% of the U.S. labor market. That share represents as much as $1.2 trillion in wages across sectors such as finance, health care, logistics, and professional services — far beyond the tech-focused roles often highlighted in automation debates.

This finding is part of a sweeping labor simulation conducted using the Iceberg Index, a modeling tool developed jointly by MIT and Oak Ridge National Laboratory. The index represents one of the most comprehensive attempts to date to visualize how AI could reshape the American workforce, not just in major innovation hubs but across all 50 states, down to specific counties and ZIP codes.

Researchers say the tool offers a detailed, forward-looking map of where disruptions are already emerging, and how those changes could ripple through local economies long before they are visible in real-world employment data.

A Digital Twin of America’s Workforce

Prasanna Balaprakash, director at Oak Ridge National Laboratory and co-leader of the project, describes the Iceberg Index as a “digital twin” of the U.S. labor market. The tool simulates the behavior and interactions of 151 million American workers, each represented as an individual agent with their own skills, tasks, occupation, and physical location.

Using this massive dataset, the index maps more than 32,000 unique skills across 923 occupations in 3,000 counties. It then evaluates which of those skills can already be performed by existing AI systems, allowing researchers to estimate exposure levels across different regions and industries.

Because the tool runs population-level experiments, it can show how tasks shift, how jobs evolve, and how workers might migrate between roles as AI adoption spreads. Balaprakash emphasized that this allows policymakers to explore possible outcomes before those changes appear in the real economy.

The Difference Between What’s Visible — and What’s Coming

One of the study’s most striking findings is the gap between the public perception of AI job risk and the actual scale of exposure.

Highly publicized layoffs in the tech sector represent only the tip of the iceberg. According to the index, the visible job disruptions — concentrated in information technology, digital services, and computing roles — account for only 2.2% of total wage exposure, or roughly $211 billion.

Beneath that visible layer lies a much larger area of vulnerability: routine work in fields such as human resources, office administration, finance, transportation logistics, and other operational roles that employ millions of workers nationwide. These sectors account for the bulk of the $1.2 trillion in wages potentially exposed to automation or augmentation from AI technologies.

The researchers caution, however, that the index is not a prediction machine and does not attempt to forecast exactly when or where job losses will occur. Instead, it is designed to provide a snapshot of what today’s AI tools could theoretically accomplish, and to help governments evaluate different policy strategies before committing resources.

States Begin Building Policies Using the Iceberg Model

MIT and ORNL have already partnered with several state governments to test and refine the platform. Tennessee, North Carolina, and Utah contributed their own workforce data to help validate the model, and they have begun using Iceberg simulations to guide workforce planning.

Tennessee became the first state to officially incorporate the findings into its policy framework, citing the Iceberg Index extensively in its newly released AI Workforce Action Plan. Utah is preparing to publish a similar report informed by Iceberg’s labor modeling.

North Carolina state senator DeAndrea Salvador, who has collaborated closely with MIT on the project, said the model stands out because it uncovers economic and workforce effects that traditional labor tools often miss. One of the platform’s strongest features, she noted, is its ability to drill down to highly localized data.

Salvador explained that policymakers can now analyze specific census blocks to determine what skills are present in a community, how likely those skills are to be automated or augmented by AI, and how those shifts could affect employment levels or regional GDP.

She added that such tools are especially important now, as states form multiple task forces and working groups to respond to rapid changes in AI technology.

AI Exposure Reaches Far Beyond Tech Hubs

The Iceberg simulations also challenge the widely held assumption that AI-related risk is concentrated mainly in coastal tech centers like California, New York, or Massachusetts. According to the model, occupations exposed to AI are distributed throughout the entire country, including inland states and rural areas that are often overlooked in discussions about technology’s economic effects.

To help address this disconnect, the research team built an interactive simulation environment that allows states to run experiments using different policy levers. Governments can explore how shifting training budgets, adjusting reskilling programs, or modifying technology adoption curves might affect local employment, income, or GDP.

The goal, according to the report, is to give policymakers a way to test interventions before committing large-scale public funds, which could total billions of dollars nationwide.

A Tool for Strengthening — Not Replacing — Local Industries

Balaprakash, who also serves on the Tennessee Artificial Intelligence Advisory Council, said the Iceberg team has shared state-specific findings with local leaders and economic development groups. Many of Tennessee’s major industries — including health care, nuclear energy, transportation, and manufacturing — still rely heavily on physical labor, which provides partial protection from pure digital automation.

The key question, he said, is how to use AI and robotics to enhance these sectors rather than undermine them. In many cases, AI tools could help improve safety, efficiency, and workforce productivity rather than directly replacing workers.

Preparing for the Future, One Scenario at a Time

For now, researchers emphasize that the Iceberg Index is still evolving. They describe it not as a finished platform, but as a flexible sandbox that states can use to stress-test future scenarios, explore possible risks, and design long-term labor strategies.

“It is really aimed towards getting in and starting to try out different scenarios,” Salvador said. As AI continues advancing, she believes tools like Iceberg will be essential for ensuring that states remain prepared — not only for potential job losses, but also for new opportunities created by emerging technologies.

By offering a granular view of AI’s impact, from the national scale down to individual ZIP codes, the Iceberg Index provides policymakers with a rare chance to anticipate the changes ahead, and to shape the future of work before it arrives.