The Iceberg Index: Measuring Skills-centered Exposure in the AI Economy

TL;DR

AI reshapes the US labor market with ripple effects that traditional metrics miss. The Iceberg Index, built on Large Population Models via AgentTorch and the Frontier supercomputer, maps skill overlap between humans and AI across 151 million workers, revealing a large hidden exposure in cognitive and administrative tasks. Visible AI adoption accounts for only 2.2% of wage value, while the broader technical exposure reaches 11.7% (~$1.2 trillion), distributed nationwide. The index provides a forward-looking, scenario-based planning tool that helps policymakers and businesses identify hotspots, reallocate training and infrastructure, and test interventions before large-scale implementation.

Abstract

Artificial Intelligence is reshaping America's \$9.4 trillion labor market, with cascading effects that extend far beyond visible technology sectors. When AI transforms quality control tasks in automotive plants, consequences spread through logistics networks, supply chains, and local service economies. Yet traditional workforce metrics cannot capture these ripple effects: they measure employment outcomes after disruption occurs, not where AI capabilities overlap with human skills before adoption crystallizes. Project Iceberg addresses this gap using Large Population Models to simulate the human-AI labor market, representing 151 million workers as autonomous agents executing over 32,000 skills and interacting with thousands of AI tools. It introduces the Iceberg Index, a skills-centered metric that measures the wage value of skills AI systems can perform within each occupation. The Index captures technical exposure, where AI can perform occupational tasks, not displacement outcomes or adoption timelines. Analysis shows that visible AI adoption concentrated in computing and technology (2.2% of wage value, approx \$211 billion) represents only the tip of the iceberg. Technical capability extends far below the surface through cognitive automation spanning administrative, financial, and professional services (11.7%, approx \$1.2 trillion). This exposure is fivefold larger and geographically distributed across all states rather than confined to coastal hubs. Traditional indicators such as GDP, income, and unemployment explain less than 5% of this skills-based variation, underscoring why new indices are needed to capture exposure in the AI economy. By simulating how these capabilities may spread under scenarios, Iceberg enables policymakers and business leaders to identify exposure hotspots, prioritize investments, and test interventions before committing billions to implementation

Figures (8)

• Figure 1: Traditional workforce metrics miss AI-mediated tasks. Census data captures jobs tied to geographic locations and business addresses. Human-AI collaboration—where workers and AI systems jointly perform tasks within occupations—creates new forms of labor that existing metrics don't capture. The Iceberg Index provides a forward-looking measure of this technical automation exposure, revealing skill-based transformation that remains invisible.
• Figure 2: Project Iceberg prepares states for the AI economy. First platform to simulate human-AI workforce interactions at national scale, enabling policymakers to assess technical exposure, test workforce strategies, and target investments before committing billions to implementation.
• Figure 3: The Iceberg Index reveals workforce exposure five times larger than visible tech adoption. Above the waterline: the Surface Index (2.2%) tracks current AI adoption in coastal technology hubs. Below the waterline: the Iceberg Index (11.7%) measures technical capability spanning administrative, financial, and professional services nationwide. This hidden mass represents where workforce preparation strategies based solely on visible tech-sector signals may fall short.
• Figure 4: Validation of skill-based framework against empirical data. (a) Skill similarity validation: Pie chart shows prediction accuracy when matching occupation pairs by skill similarity to actual career transitions from O*NET data. For pairs our framework identifies as highly similar (>80% similarity score, dark green), 85% correspond to observed career transitions, confirming skill embeddings capture genuine labor market structure. Lower similarity thresholds (70-80%; <70%) show decreasing alignment with actual transitions. (b) Adoption validation: State-level comparison shows 69% agreement between Anthropic Economic Index (actual AI usage patterns) and Surface Index (predicted technical exposure). Strong alignment at both extremes: 8 of 13 leading states and 9 of 13 aspiring states match perfectly, confirming the Index identifies structural exposure before widespread adoption.
• Figure 5: Surface Index validation: Current technology-sector exposure across US states. The average surface index of the country is 2.2, maximum is 4.2 (Washington) and minimum is 0.2 (Guam). 18 states (or territories) are above the national average and 36 are below the national average.