AI Vulnerability Score™ Methodology
Full transparency on how we calculate your score. Every input, weight, transformation, and data source is documented here.
What This Score Measures
The AI Vulnerability Score is a 0–100 composite index that measures how well you can absorb an AI-driven income disruption. It uniquely combines occupation-level AI displacement exposure with household balance-sheet resilience into a single, actionable number.
The score architecture follows a threat (20%) – buffer (53%) – adaptability (27%) hierarchy modeled after actuarial risk assessment, grounded in FICO methodology principles.
Six Input Variables
Your specific occupation scored using Karpathy's AI Exposure analysis of 342 BLS occupations (theoretical risk), blended with Anthropic's real-world adoption velocity data (how fast AI is actually being deployed in your field). This creates a composite risk score that captures both what AI could automate and what it currently is automating.
Liquid savings divided by monthly essential expenses. The strongest single predictor of surviving any income shock per CFPB, Vanguard, and Financial Resilience Institute research.
Total monthly debt obligations divided by gross monthly income. High DTI accelerates runway depletion because fixed debt payments consume replacement income.
How applicable your current skills are to roles outside your current occupation. Brookings found 6.1 million workers face both high AI exposure AND low adaptive capacity.
Percentage of total household income not dependent on primary employment. Even modest diversification meaningfully reduces vulnerability.
Liquid assets divided by total assets. Captures wealth accessibility beyond runway — a person with $500K in home equity but $2K cash is more vulnerable than someone with less total wealth but more liquidity.
Non-Linear Scoring
Each raw input is transformed through a variable-specific non-linear function before weighting. Linear scoring would misrepresent the demonstrably non-linear relationship between these variables and resilience.
For example, financial runway uses a concave (square root) transformation — going from 0 to 3 months of savings produces a much larger score improvement than going from 12 to 15 months. This reflects research showing even small emergency savings buffers dramatically reduce financial distress.
Interaction Terms
High Risk × Low Skill: When occupation risk is high AND skill transferability is low, a penalty is applied. This reflects Brookings' finding that this combination creates qualitatively different risk.
Runway × DTI: High DTI accelerates runway depletion. Effective runway is adjusted downward based on DTI level, reflecting the reality that debt payments consume replacement income during disruption.
Score Range and Calibration
The score uses a 15–95 effective range within the 0–100 scale, following FICO methodology. A floor of 15 prevents nihilistic interpretations. A ceiling of 95 maintains aspiration.
Blended Occupation Risk Score
The occupation risk dimension (20% of total score) combines two data sources into a single risk metric:
70% — Karpathy AI Exposure Score (Theoretical Risk). 342 BLS occupations scored 0–10 based on how much AI could reshape each role. Jobs done entirely on a computer score highest (8–10); jobs requiring physical presence in unpredictable environments score lowest (0–2). Source: Andrej Karpathy, "AI Exposure of the US Job Market" (karpathy.ai/jobs, March 2026).
30% — Anthropic Adoption Velocity (Real-World Risk). The ratio of observed AI task coverage to theoretical coverage in each occupational category. A high-exposure job where AI adoption is already rapid (like computer programming at 75% observed coverage) scores higher than one where adoption is still slow despite high theoretical potential. Source: Massenkoff & McCrory, "Labor Market Impacts of AI" (Anthropic Research, March 2026).
The gap between theoretical and observed exposure represents your preparation window — the time before theoretical automation potential becomes real-world displacement. As of March 2026, actual AI adoption is a fraction of what's theoretically possible in most occupations. This gap is expected to narrow over the next 2–5 years.
Important Limitations
AI displacement predictions carry significant uncertainty. LLM-based scoring of jobs for LLM replaceability is methodologically circular. Exposure does not equal displacement — a high score means high task overlap, not guaranteed job loss. Anthropic's own research shows no systematic increase in unemployment for highly exposed workers since late 2022, though there is suggestive evidence of slowed hiring for younger workers in exposed occupations.
The occupation risk scores are based on current research (March 2026 snapshot) but may shift as AI capabilities evolve. The weights are literature-informed but not regression-validated against actual displacement outcomes.
This tool is one input among many for understanding your financial resilience. It is educational content, not personalized financial advice. Consult a licensed financial professional for individualized guidance.
Data Sources
Occupation AI Exposure
- Karpathy, A. (2026). AI Exposure of the US Job Market. karpathy.ai/jobs — 342 BLS occupations scored 0–10
- Massenkoff, M. & McCrory, P. (2026). Labor market impacts of AI: A new measure and early evidence. Anthropic Research.
- Eloundou et al., "GPTs are GPTs" (Science, 2024)
- BLS, Occupational Outlook Handbook — employment, pay, education, and growth projections
Financial Resilience
- Federal Reserve, Survey of Consumer Finances (2022)
- Federal Reserve, SHED Survey (2024)
- BLS, Displaced Workers Survey (2024)
- Vanguard, How America Saves (2025)
- CFPB, Financial Resilience Research
- Brookings Institution, Adaptive Capacity Research (2019–2026)
- Goldman Sachs, AI and Economic Growth Research (2023–2025)
- World Economic Forum, Future of Jobs Report (2025)
- ILO, Generative AI and Jobs Working Papers (2023–2025)
© 2026 FinanciallyOptimized.com | For educational purposes only. Not financial advice.