Anthropic Economic Index — Job Exposure
Anthropic · 2025
Used for:Primary AI-exposure metric (observed_exposure) per SOC occupation — replaces Frey-Osborne 2013
View datasetFutureGrid uses current, authoritative datasets to power every metric on this site.
These datasets are directly loaded and used to compute metrics.
Anthropic · 2025
Used for:Primary AI-exposure metric (observed_exposure) per SOC occupation — replaces Frey-Osborne 2013
View datasetAnthropic / O*NET / BLS (bundled) · 2025
Used for:Median salary fallback, job forecast, job zone, bright outlook, sector (JobFamily)
View datasetU.S. Bureau of Labor Statistics · 2025
Used for:per-occupation employment + median wage (authoritative; overrides AEI bundle when BLS_API_KEY set); employmentHistory and wageHistory annual series 2016–2025
View datasetU.S. Bureau of Labor Statistics · 2024
Used for:SOC-keyed 2024–2034 projected employment, employment change, growth rates, wages, and annual openings in employment-projections.json
View datasetAnthropic / BLS (bundled) · 2023
Used for:Employment totals by occupation (legacy fallback — superseded by OEWS API when key is set)
View datasetAnthropic · 2025
Used for:Per-country AI usage index, usage %, usage count, GDP per working-age capita
View datasetWorld Bank · 2024
Used for:Supplemental China GDP per working-age capita when Anthropic country usage metrics are not reported
View datasetEurostat · 2024
Used for:Supplemental multi-country enterprise AI adoption rate for European reporting countries
View datasetOECD · 2025
Used for:Supplemental enterprise AI adoption and individual generative-AI usage rates by reporting country
View datasetU.S. Census Bureau · 2018
Used for:Supplemental U.S. employer-firm artificial-intelligence total-use counts and percentages
View datasetCNNIC / Xinhua / State Council of the People's Republic of China · 2025
Used for:Supplemental China generative-AI user count, penetration rate, and six-month user growth proxy
View datasetQuestMobile · 2025
Used for:Supplemental China AI application MAU, category scale, and token-volume proxy metrics
View datasetGuancha, citing QuestMobile · 2026
Used for:Secondary-source China-native AI app MAU figures for Doubao and DeepSeek
View datasetHugging Face · 2026
Used for:Supplemental open-model download and like-count proxies for Qwen, DeepSeek, Kimi, ChatGLM, and Baichuan
View datasetGitHub · 2026
Used for:Supplemental developer-ecosystem proxy metrics for AI repositories
View datasetStack Overflow / R4DS TidyTuesday · 2024
Used for:Supplemental developer AI-tool usage, AI sentiment, and respondent-country proxy metrics
View datasetWorld Bank Data360 / OECD.AI · 2025
Used for:Supplemental country-level AI research activity proxy from AI publication totals
View datasetU.S. Department of Labor / Employment and Training Administration · 2024
Used for:Top skills (by importance) per SOC occupation
View datasetU.S. Department of Labor / O*NET Resource Center · 2026
Used for:Build-time enrichment for occupation descriptions, tasks, detailed work activities, technology skills, and related occupations
View datasetLuke Duncalfe (GitHub) · 2024
Used for:Numeric ISO 3166-1 → alpha-3 mapping used to annotate world-atlas country features
View datasetIndeed Hiring Lab · 2026
Used for:Share of AI and generative-AI job postings as a % of all postings (7-day trailing average); country-level time series in ai-demand.json
View datasetChallenger, Gray & Christmas, Inc. · 2026
Used for:US employer-announced job cuts citing AI as stated reason (monthly and annual totals); ai-layoffs.json — ⚠️ LICENSE RISK: data is proprietary; see data/COMPLIANCE.md
View datasetEd Felten, Manav Raj, Robert Seamans · 2021
Used for:Per-SOC AIOE exposure scores (application-ability relatedness) as a supplemental AI-exposure metric; aioe-exposure.json — ⚠️ LICENSE RISK: no open license; see data/COMPLIANCE.md
View datasetMicrosoft AI Economy Institute · 2026
Used for:GenAI diffusion % of working-age population across three periods (H1 2025 / H2 2025 / Q1 2026) for 147 economies; global-ai-metrics.json
View datasetInternational Monetary Fund · 2023
Used for:Overall AI readiness composite (0–1, 178 countries) and four AIPI sub-indices (DI, HCLMP, IEI, RE); global-ai-metrics.json — ⚠️ CONDITIONAL: non-commercial only; see data/COMPLIANCE.md
View datasetOxford Insights · 2023
Used for:Government AI readiness composite score 0–100 for 193 countries; global-ai-metrics.json
View datasetOpenRouter · 2026
Used for:Provider/model catalog snapshot, endpoint-provider counts, and country-level catalog-proxy mapping; openrouter-models.json — catalog footprint only, not usage, traffic, demand, or deployment geography
View datasetU.S. Bureau of Labor Statistics · 2026
Used for:Monthly job openings, hires, quits, layoffs & separations by industry (JOLTS); jolts.json
View datasetOpenAI · 2023
Used for:Per-SOC GPT-4 exposure ratings (E1/E2/E3 tiers, dv_rating + human annotation); llm-exposure.json
View datasetYahoo Finance / Verizon Media · 2026
Used for:Monthly closing prices for AI-sector ETF tickers (XLC, XLK, BOTZ, ARKQ, etc.) used to derive AI market signals; market-ai-signals.json — ⚠️ HIGH LICENSE RISK: Yahoo ToS prohibits redistribution; see data/COMPLIANCE.md
View datasetFutureGrid / Yahoo Finance / Alpha Vantage · 2026
Used for:Historical adjusted-close observations for public AI-company watchlist; ai-company-stocks.json — descriptive market history only, not investment advice, forecasts, recommendations, or redistribution-cleared market data
View datasetU.S. Bureau of Labor Statistics · 2026
Used for:State-level monthly unemployment rate, employment, and labor force totals; state-labor.json
View datasetU.S. Bureau of Labor Statistics · 2025
Used for:State-level annual employment and wages by industry (NAICS) from QCEW; state-qcew.json
View datasetCalifornia Employment Development Department (EDD) · 2026
Used for:Mass layoff and plant closure notices filed with California EDD under the WARN Act; warn-notices.json
View datasetU.S. DOL OFLC · 2025
Used for:Certified H-1B LCA volume, offered-wage percentiles, top occupations/employers/states over FY2016–FY2025; h1b-trends.json
View datasetFutureGrid · 2025
Used for:SOC-keyed 2016–2025 deterministic job-postings seed for correlation experiments until a licensed historical postings provider is wired in
View datasetU.S. Bureau of Labor Statistics · 2025
Used for:SOC-keyed occupational requirements concepts for occupational-requirements.json: preparation burden, physical presence, work conditions, decision-making, and FutureGrid automation-friction score
View datasetInternational Labour Organization (ILO) · 2025
Used for:International occupation-mix shares by ISCO-08 major group (1-9) for included countries in international-occupation-mix.json
View datasetEpoch AI · 2026
Used for:AI training-compute/cost/power frontier trends (AI Frontier page)
View datasetThese publications inform interpretation and provide independent validation context — they are not directly loaded into the application.
International Monetary Fund · 2024 · Context/validation reference only — not directly loaded
OECD · 2023 · Context/validation reference only — not directly loaded
International Labour Organization · 2025 · Context/validation reference only — not directly loaded
Natural Earth / Mike Bostock (world-atlas npm package) · 2023 · World map choropleth geometry — country polygons at 110m resolution, Antarctica dropped; numeric ISO ids mapped to ISO 3166-1 alpha-3 via lukes/ISO-3166-Countries-with-Regional-Codes crosswalk. AEI usageIndex primary choropleth value; proxy context shown for restricted countries (e.g. China).
Carl Benedikt Frey & Michael A. Osborne, Oxford Martin School · 2013 · Historical automation probability per SOC (702 occupations) as a context/comparison baseline; automation-baseline.json — ⚠️ LICENSE RISK: no open license; see data/COMPLIANCE.md
Technical Note
automationRisk bands are percentile-calibrated from the aiExposure distribution (756 occupations): Very High = top ~8% (aiExposure > 0.3104), High = next ~12% (> 0.1378), Medium = next ~25% (> 0.0080), Low = remainder (≤ 0.0080). aiExposure = observed_exposure from Anthropic Economic Index (Claude AI-usage based, not Frey-Osborne 2013). employment: real OEWS 2025 figures (755 occupations updated via BLS Public Data API). medianSalary: OEWS 2025 where available (752 updated), AEI-bundled wage otherwise. employmentHistory/wageHistory: OEWS multi-year annual series 2019–2025 — 2025 via BLS Public Data API (755 occ); 2019–2023 via archived national Excel flat files (Wayback Machine); note that BLS OEWS OEUN API series stores only the current-vintage snapshot so older years require flat-file download. growthRate is null (no authoritative per-SOC % growth in AEI files). projectedOpenings from wage_data.JobForecast (BLS-EP annual openings) where > 0. China is included as a supplemental country row with World Bank 2024 GDP per working-age capita; Anthropic Claude.ai usage metrics for China are not reported and remain null. O*NET skills: successfully loaded
Export the underlying datasets as CSV or JSON for offline analysis. Generated client-side — no server request.
Occupations
Automation risk, probability, salary, employment, and outlook for all occupations. CSV is flat (skills and history arrays excluded).
Countries
AI usage index, diffusion rates, readiness scores, and government AI readiness by country. CSV is flat (nested sub-indices excluded).
Data Sources
Full citation list with publishers, years, licenses, and URLs. JSON includes top-level metadata.
JSON preserves all fields including nested arrays and objects. CSV is flat — nested structures are excluded.