AI's Iceberg Index: What HR Needs to Do About the 11.7% Exposure
A new study from researchers at MIT and Oak Ridge National Laboratory finds that current AI tools can already perform tasks tied to 11.7% of total U.S. wage value - about $1.2 trillion. That exposure spans finance, healthcare, administrative services, and professional roles.
Key point for HR: this is technical exposure, not a forecast of job loss. It's a map of where AI can match human skills today, so you can plan workforce moves before adoption scales.
What the Iceberg Index Measures
The team modeled the labor market with 32,000 skills, 923 occupations, and 3,000 counties, representing 151 million workers. They applied the same skill taxonomy to 13,000+ AI tools (copilots, workflow systems, and more) to find overlap between what people do and what AI can already do in practice.
The study also defines a narrower Surface Index - the visible adoption cluster in computing and tech roles - at 2.2% of wage value (about $211 billion). The broader Iceberg Index shows the bigger, less visible share of tasks across non-tech roles that are already within reach of current tools.
Where Exposure Is Concentrated
- Administrative and support: document prep, scheduling, reporting, templated communications.
- Finance and professional services: analysis, reconciliation, compliance checks, drafting client materials.
- Healthcare ops (non-clinical): intake documentation, prior auth packets, coding assistance, panel reports.
Exposure shows up in every state - even where tech employment is small. Manufacturing regions see higher-than-expected exposure in white-collar coordination and support functions.
Why This Matters for HR
AI will first shift tasks, then job mix. Roles won't vanish overnight, but task portfolios will. HR teams that get ahead with task-level planning, skills data, and targeted training will cut costs, protect productivity, and reduce disruption.
Signals from States Already Testing the Model
Tennessee, North Carolina, and Utah partnered with the researchers to test accuracy and explore policy options. The platform lets officials analyze county-level skill patterns and trial training or investment scenarios before committing funds.
For HR, treat that as a cue: run local, skill-level planning before you roll out tools or restructure roles.
90-Day HR Action Plan
- Weeks 1-2: Task audit - For target functions (admin, finance, operations), list top recurring tasks and time spent. Mark tasks that are rules-based, document-heavy, or template-driven.
- Weeks 3-4: Pilot candidates - Pick 2-3 high-volume tasks (e.g., report drafting, invoice checks, scheduling). Define success metrics: cycle time, error rate, rework, cost per task.
- Weeks 5-8: Tool trials - Test 1-2 AI copilots or workflow tools per task with a small user group. Keep human review steps. Track results weekly.
- Weeks 9-12: Scale or stop - Standardize what works (SOPs, prompts, QA). Update job descriptions with new task splits. If gains are weak, exit fast and pick the next task.
Job Architecture and Skills
- Rewrite job descriptions at the task level. Separate "AI-assist" from "human-only" responsibilities.
- Shift to skills-based HR: tag roles with the underlying skills (communication, compliance review, data analysis, documentation synthesis).
- Create new career paths that blend domain expertise with AI supervision and quality control.
Guardrails and Change Management
- Data use: define what data can feed AI tools; log prompts and outputs for audits.
- Human-in-the-loop: mandate human review for anything customer-facing, financial, legal, or clinical.
- Bias and quality checks: spot-check outputs weekly; keep a feedback loop to improve prompts and SOPs.
- Performance: tie incentives to quality and cycle-time improvements, not tool usage.
Metrics That Matter
- Cycle time per task (before vs. after pilot)
- First-pass yield and rework hours
- Cost per task and cost per error
- Time-to-proficiency for employees using new tools
What to Communicate to Employees
- AI targets tasks, not people. The goal is to remove low-value work and upskill the team.
- Every pilot has quality checks and human review.
- Training is available, with clear paths to higher-value responsibilities.
Training Resources
If you're building a skills-first program and need practical curricula:
- Courses by Job - map training to role families (HR, finance, operations).
- Courses by Skill - upskill on prompts, data analysis, and workflow design.
Further Reading
Bottom Line for HR
The Iceberg Index shows that a meaningful slice of work is already within the reach of current AI tools. Start with task audits, run tight pilots, bake in guardrails, and retrain for the skills that rise in value.
The faster you build this loop, the smoother your team transitions as adoption spreads across functions and geographies.
Your membership also unlocks:
