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

**Authors:** Ayush Chopra, Santanu Bhattacharya, DeAndrea Salvador, Ayan Paul, Teddy Wright, Aditi Garg, Feroz Ahmad, Alice C. Schwarze, Ramesh Raskar, Prasanna Balaprakash  
**arXiv:** [2510.25137](https://arxiv.org/abs/2510.25137)  
**Date:** 2025-10-29  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2510.25137v2/

## Key points

- Artificial Intelligence is revolutionizing America's \$9.4 trillion labor market across various industries, not just technology sectors
- Transformation in quality control tasks within automotive plants has ripple effects through logistics networks, supply chains, and local service economies
- Project Iceberg uses Large Population Models to simulate the human-AI labor market, representing 151 million workers and over 32,000 skills interacting with AI tools
- Iceberg Index quantifies the wage value of skills that AI systems can execute within each occupation, focusing on technical exposure rather than displacement outcomes or adoption timelines
- Visible AI adoption concentrated in computing and technology sectors (\$211 billion), but extends deeper into cognitive automation across administrative, financial, and professional services (\$1.2 trillion)
- Extensive exposure geographically dispersed across all states rather than limited to coastal hubs
- Traditional economic indicators explain less than 5% of skills-based variation in the AI economy, highlighting the need for new indices
- Iceberg simulation empowers policymakers and business leaders to identify exposure hotspots, prioritize investments strategically, and test interventions effectively before implementation
- Technology-intensive states like California and Washington exhibit high values in both visible AI adoption and deep-rooted technical capabilities based on their unique composition

## Summary

Artificial Intelligence is revolutionizing America's \$9.4 trillion labor market by extending its effects across various industries and not just visible technology sectors. The transformation brought about by AI in quality control tasks within automotive plants has far-reaching consequences that ripple through logistics networks, supply chains, and local service economies. To accurately measure these ripple effects, traditional metrics used to assess workforce outcomes are inadequate as they only consider employment results post-disruption instead of identifying the overlap between AI capabilities and human skills before adoption solidifies. This gap is bridged by Project Iceberg which utilizes Large Population Models to simulate the human-AI labor market. It represents 151 million workers as autonomous agents proficient in over 32,000 skills interacting with thousands of AI tools. The introduction of the Iceberg Index provides a skills-centered metric that quantifies the wage value of skills that AI systems can execute within each occupation. This index focuses on technical exposure where AI can perform occupational tasks rather than predicting displacement outcomes or adoption timelines. Analysis reveals that while visible AI adoption is concentrated in computing and technology sectors (representing 2.2% of wage value amounting to approximately \$211 billion), the technical capability extends much deeper below the surface through cognitive automation spanning administrative, financial, and professional services (comprising 11.7% equating to around \$1.2 trillion). This extensive exposure is geographically dispersed across all states rather than being limited to coastal hubs. Traditional economic indicators like GDP, income levels, and unemployment rates explain less than 5% of this skills-based variation, underscoring the necessity for new indices capable of capturing exposure within the AI economy accurately. By simulating potential scenarios depicting how these capabilities may proliferate, Iceberg empowers policymakers and business leaders to identify exposure hotspots, prioritize investments strategically, and test interventions effectively before committing substantial resources towards implementation. Furthermore, technology-intensive states such as California and Washington exhibit high values on both surface-level visible AI adoption as well as deep-rooted technical capabilities depending on their unique composition. Through meticulous mapping of human workforce skills and understanding where they intersect with AI capabilities lies the key to navigating the evolving landscape of work in an increasingly automated world.