MIT ‘Iceberg Index’ maps hidden AI risk to 12% of U.S. jobs
MASSACHUSETTS, UNITED STATES — A new study from Project Iceberg, developed by the Massachusetts Institute of Technology (MIT) in collaboration with Oak Ridge National Laboratory, reveals that the visible disruption of artificial intelligence (AI) on tech jobs is just a fraction of its potential impact.
The “Iceberg Index” estimates that AI’s technical capabilities extend to cognitive and administrative work, affecting 11.7% of the United States’ labor market’s wage value.
This model, they say, creates a “digital twin” of the U.S. labor market, simulating 151 million workers and mapping their skills against current AI capabilities to quantify the immediate risk of automation across the economy, according to a CNBC report.
“Project Iceberg enables policymakers and business leaders to identify exposure hotspots, prioritize training and infrastructure investments, and test interventions before committing billions to implementation,” the report notes.
AI’s hidden white‑collar automation risk
Current workforce metrics and headlines focus on visible AI adoption in technology sectors, which represent a small portion of the economy.
Project Iceberg’s “Surface Index” quantifies this visible exposure at just 2.2% of national wage value, approximately $211 billion, concentrated in coastal hubs like Washington (4.2%), California (3.0%), and Virginia (3.6%), among 1.9 million technology workers.
This aligns with real-world adoption data but captures only jobs in software development, data science, and related fields where AI tools are actively accelerating work.
Beneath this surface, the full “Iceberg Index” reveals a far larger sphere of technical capability. The Index measures a national average of 11.7% exposure—about $1.2 trillion in wage value—spanning administrative, financial, and professional service occupations where AI has demonstrated skill overlap.
The report notes, “Administrative and financial tasks where AI demonstrates capability span five times more wage value than visible tech disruption—and are geographically distributed across all states, not just coastal.”
This hidden cognitive automation is geographically distributed across all states, not confined to tech hubs, indicating that traditional economic indicators like GDP and unemployment explain less than 5% of this skills-based variation and fail to capture the broader transformation.
Industrial U.S. states face unexpected AI exposure
The research identifies a significant automation surprise for states that may underestimate their risk by focusing only on visible tech-sector signals. Industrial heartland states like Ohio (11.8%) and Tennessee (11.6%) show modest Surface Index values but double-digit Iceberg Index exposure.
This gap is driven by AI’s capabilities in white-collar functions—such as financial analysis, administrative coordination, and professional services—that support manufacturing and logistics, creating a blind spot for policymakers primarily focused on physical automation.
The structure of exposure—whether concentrated or distributed across industries—dictates vastly different policy responses. Analysis using the Herfindahl-Hirschman Index shows Northeastern states often have exposure concentrated in finance and technology, enabling targeted action.
In contrast, Manufacturing Belt states exhibit distributed exposure across logistics, production, and administration, requiring broad, multi-sector coordination.
This illustrates the importance of tools like Project Iceberg’s simulation sandbox, which allows states to test interventions and align billion-dollar investments in training and infrastructure with their specific exposure profile before adoption crystallizes.
The Iceberg Index ultimately reveals that the true disruption of AI lies not in the visible tech sector, but in its pervasive, geographically dispersed capacity to automate cognitive work, demanding a fundamental shift in policy from reactive measurement to proactive, simulation-driven intervention.
