Credits & Data Sources

AI Changing Work draws on publicly available research, open classification standards, and open-source technologies. This page credits the references and tools behind the site. The AI exposure figures published here are drawn from the datasets listed below, and the employment, wage and projection figures are transcribed from U.S. Bureau of Labor Statistics Employment Projections. Source values are carried at the value each publisher assigned: AI Changing Work does not rescale, reweight, reclassify or blend them, and calculates no exposure score of its own. AI Changing Work performs the mapping onto O*NET occupation and task codes. Presentation is localised — for example, where a publisher expresses a value as a worded category rather than a number, the wording shown is that category translated into the page's language. Localisation changes how a value is written, not which value its publisher assigned.

Data Sources

O*NET OnLine— U.S. Department of Labor

Source of the occupational classification taxonomy — the codes and titles used to organize the occupations listed here, and the task and occupation codes the external exposure datasets are mapped onto.

CC BY 4.0
Occupational Outlook Handbook— U.S. Bureau of Labor Statistics

Source of the employment, wage and projected-change figures shown on this site, transcribed from BLS Employment Projections 2025-35 at the SOC occupational-group level.

Public Domain
Employment Projections — AI exposure categories— U.S. Bureau of Labor Statistics

Source of the relative AI exposure band shown on occupation pages, carried from the 2025-35 edition at the value BLS assigned. Presentation is localised — the band is a worded category rather than a number, and the wording shown is that category translated into the page's language. Localisation changes how a value is written, not which value its publisher assigned. The band is a relative rank, not an absolute level, and it is not a first-hand measurement: BLS builds it by grouping each occupation's percentile ranks across several published studies into four bands, and two of those studies are also credited on this page.

Public Domain

Source of the observed exposure and task penetration figures shown on this site, reproduced without modification under CC BY 4.0.

HAI AI Index— Stanford University

Referenced for global AI adoption rates, investment data, and policy analysis.

OECD— Organisation for Economic Co-operation and Development

Referenced for labor statistics across 38 countries and the AI Policy Observatory.

ILO— International Labour Organization

Referenced for global employment trends and the GenAI occupational exposure index.

Research & Publications

Key institutions whose research is cited in our blog content and analysis:

Technology

Open-source technologies used to build this site:

ProjectRoleLicense
Next.jsReact frameworkMIT License
PostgreSQLDatabasePostgreSQL License
Tailwind CSSStylingMIT License
shadcn/uiUI componentsMIT License
RechartsData visualizationMIT License
next-intlInternationalizationMIT License
LucideIconsISC License

AI Tools

Data analysis, content generation, and translations are performed with the assistance of Anthropic Claude.

See our AI Content Disclosure for details →

Open Data Licenses

We comply with the following data-use terms:

  • This site includes information from O*NET Resource Center by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified some of this information — occupation and task statements are machine-translated into other languages and mapped onto this site's occupation set. USDOL/ETA has not approved, endorsed, or tested these modifications. View terms
  • BLS data is in the U.S. Government Public Domain.
  • OECD data is used in accordance with the OECD Terms and Conditions. View terms
  • Occupational classification vocabulary comes from the following sources: ILO ISCO-08, International Labour Organization (no licence stated by the ILO for this publication; reused with attribution); O*NET® 30.2, U.S. DOL/ETA (CC BY 4.0); the KSCO 8th–ISCO-08 Correspondence Table published by the Korea Data Agency under KOGL Type 3 (Attribution – No Derivatives), quoted verbatim; BSCO-2020 by the Bangladesh Bureau of Statistics (BBS), reproduced without modification under no open licence and not endorsed by BBS or the ILO; 総務省「国際標準職業分類(ISCO)2008年改定版(仮訳)分類項目名一覧(和英対照)」— 原著:ILO ISCO-08 © International Labour Organization(日本語訳は総務省がILOの許可を得て作成); ESCO API 2026-08-08 (bridge v1.2.1), European Commission, DG EMPL (© European Union; reused under the European Commission reuse policy, CC BY 4.0). AI Changing Work has modified this material: ESCO records were mapped to ISCO-08 groups, selected by locale and restructured into this site's occupation schema. The ESCO texts themselves are reproduced without alteration, except that character-encoding errors in some source texts have been repaired. The European Commission has not approved, endorsed or verified these modifications. To request removal of any of this material, contact us. Request removal
  • Anthropic Economic Index (labor_market_impacts) is used under CC BY 4.0. Attribution: Data: Anthropic Economic Index — labor_market_impacts, CC-BY, https://huggingface.co/datasets/Anthropic/EconomicIndex View terms
  • AI Changing Work uses Anthropic Economic Index data under CC-BY; Anthropic does not endorse or sponsor this site or its analyses.
  • The OpenAI "GPTs are GPTs" exposure rubric is used under the MIT License, Copyright (c) 2024 OpenAI. View terms
  • U.S. Bureau of Labor Statistics Employment Projections 2025-35 are a work of the U.S. federal government and are in the public domain (17 U.S.C. 105).

Task-level exposure

This page includes information from the O*NET® 20.1 Database, O*NET® 27.2 Database and O*NET® 30.2 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. AI Changing Work has modified some of this information (Korean translations of task statements). USDOL/ETA has not approved, endorsed, or tested these modifications.

Observation data: Anthropic Economic Index (CC BY 4.0).

Prediction data: Eloundou et al., "GPTs are GPTs" (2023) — task labelset on the O*NET 27.2 basis.

MIT License · Copyright (c) 2024 OpenAI

MIT License — OpenAI GPTs-are-GPTs
MIT License

Copyright (c) 2024 OpenAI

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Country Labour-Market Data & Sources

AI exposure index: The ILO exposure figures on this site are reproduced without modification from "Generative AI and Jobs: A Refined Global Index of Occupational Exposure" (ILO Working Paper 140, Gmyrek et al., 2025, © ILO 2025), published by the International Labour Organization under CC BY 4.0. AI Changing Work performs the mapping of ISCO-08 4-digit unit-group values onto this site's occupations; the ILO's figures themselves are unmodified. ILO Working Paper 140

Gmyrek, P., Berg, J., Kamiński, K., Konopczyński, F., Ładna, A., Nafradi, B., Rosłaniec, K., Troszyński, M. Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO Working Paper 140. Geneva: International Labour Office, 2025. © ILO.

AI Changing Work uses ILO material (the Working Paper 140 exposure index) under CC BY 4.0; the ILO does not endorse or sponsor this site or its analyses.

International Labour Organization (ILO) — ISCO-08 Structure

Korean occupation titles are quoted verbatim from the KSCO 8th – ISCO-08 Correspondence Table, published by the Korea Data Agency (Statistical Classification Portal) and released under KOGL Type 3 (Attribution – No Derivatives). Source: https://kssc.mods.go.kr. No modifications were made. Not endorsed by, and not affiliated with, the Korea Data Agency.

The KSCO–KECO correspondences shown here are based on the crosswalk in Appendix 4 of the "Korean Employment and Occupational Classification (KECO) 2025 Explanatory Manual" (Ministry of Employment and Labor & Korea Employment Information Service, 2025-04-25), displayed without modification. KECO codes and titles follow Ministry of Employment and Labor Notification No. 2024-63 (KECO 2025).

Korean wages & job outlook: The Korean per-occupation occupation titles, wage levels (lower / average / upper), job outlook, job satisfaction, duties, related occupations, work abilities, knowledge, work environment, personality, interests, work values, work activities (importance and level), related majors, related certifications, and how-to-enter information on this site use 'Worknet Occupational Information (KNOW)', created by the Korea Employment Information Service and opened under KOGL Type 4 (Source Indication – Commercial Use Prohibited – No Derivative Works). The work is available free of charge from the Open Data Portal (data.go.kr). Text is reproduced verbatim and figures are unchanged in value (wages are abbreviated with K for 10 million KRW), and this site uses the data on a non-commercial basis. Which KNOW occupation appears on which occupation page of this site is determined by this site, not by a correspondence defined by the Korea Employment Information Service. Which items are shown may differ from one occupation page to another. (Retrieved: 2026-08-17)

French occupational information & labour-market statistics: The French occupation titles, role definitions and competencies on this site use ROME 4.0 (version 61), opened by France Travail under the Licence Ouverte / Open Licence 2.0; the wage and hiring-tension statistics use the France Travail "Marché du travail" API (which states compliance with the Etalab licence 2.0). The source data is available free of charge from data.gouv.fr and the France Travail developer portal. The figures are displayed without modification. (Source last updated: ROME 4.0 version 61 2026-06-15 · Marché du travail API 2026-06-12 / retrieved: 2026-08-17) France Travail does not endorse or sponsor this site.

Korea Occupation Direct Crosswalk v1.1

This dataset includes information from the O*NET® 31.0 Database (https://www.onetcenter.org/database.html) by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified all or some of this information: O*NET-SOC codes and titles are reproduced unchanged and paired with KSCO-8 unit groups by AI Changing Work's judgment; relation labels, statuses and the ACW-KR-* nodes are AI Changing Work's additions. USDOL/ETA has not approved, endorsed, or tested these modifications.

KSCO 코드·명칭: 한국표준직업분류(제8차 개정) — 통계청 고시 제2024-328호 (2024-07-01 고시, 2025-01-01 시행). 저작권법 제7조 제2호의 고시 항목이다. 명칭 표기(가운뎃점·띄어쓰기)는 해설서 2차 정오 반영판의 표기를 따랐으며, 고시 항목표와는 18개 명칭에서 가운뎃점 글리프나 띄어쓰기만 다르다. 통계청은 2025년 10월 국가데이터처로 개편되었다. 이 표는 통계청·국가데이터처의 공식 연계표가 아니다.

Occupation master (O*NET® 31.0 Database)

The occupation pages (/jobs) include information from the O*NET® 31.0 Database (https://www.onetcenter.org/database.html) by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA); task tables on these pages also include information from one or more of the O*NET® 20.1, 27.2 and 30.2 Databases and use the O*NET-SOC 2010 to 2019 Crosswalk (https://www.onetcenter.org/taxonomy/2019/walk.html) by USDOL/ETA. Used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified all or some of this information: O*NET-SOC codes, titles and task statements are reproduced in English without change; task-type labels are shown in the page's language and tasks are listed core first; in task tables, task statements are shown in the wording of the O*NET version each source used, next to values published by third parties, and task statements of O*NET-SOC 2010 occupations are shown on the occupations the crosswalk links them to; any Korean occupation titles and task statements shown on Korean-language pages are AI Changing Work's translations; any KSCO-8 unit groups or ROME 4.0 job profiles linked to an occupation were paired with it by AI Changing Work's judgment, and the relation labels and statuses are AI Changing Work's additions; the occupations with AI Changing Work codes, and any O*NET-SOC occupations listed as nearest to them, are AI Changing Work's additions and selections and are not part of the O*NET-SOC taxonomy. USDOL/ETA has not approved, endorsed, or tested these modifications.

This page includes information from the O*NET® 31.0 Database (https://www.onetcenter.org/database.html) by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified all or some of this information: O*NET-SOC codes and titles are reproduced in English without change; any Korean occupation titles shown on the Korean-language page are AI Changing Work's translations; the occupations with AI Changing Work codes are AI Changing Work's additions and are not part of the O*NET-SOC taxonomy. USDOL/ETA has not approved, endorsed, or tested these modifications.

AI exposure figures on these pages are published by third parties and credited on this page: the OpenAI "GPTs are GPTs" exposure rubric (MIT License) and the Anthropic Economic Index (CC-BY); ILO Working Paper 140 (CC BY 4.0); and the U.S. Bureau of Labor Statistics AI exposure categories (public domain). ILO values, published for ISCO-08 unit groups, are linked to O*NET occupations by applying the ISCO-08 to 2010 SOC crosswalk and the 2010 SOC to 2018 SOC crosswalk of the U.S. Bureau of Labor Statistics (public domain) as published, without case-by-case selection; many of their matches are partial. Values published per SOC code are shown on every O*NET occupation with that code. Task statements of O*NET-SOC 2010 occupations in the task tables are linked to O*NET occupations by applying the O*NET-SOC 2010 to 2019 Crosswalk of USDOL/ETA (CC BY 4.0) as published, without case-by-case selection.

ISCO-08 unit group titles and texts from the International Labour Organization and ESCO occupations from the European Commission are also shown on these pages and are credited under Licenses on this page. The ISCO-08 unit groups shown with ILO texts are linked to O*NET occupations by the same U.S. Bureau of Labor Statistics crosswalks. The ESCO occupations shown are linked to O*NET occupations by the ESCO-O*NET Crosswalk, version 1 (September 2022), which pairs O*NET-SOC 2019 with ESCO v1.1.0, by the European Commission (ESCO Secretariat) and the U.S. Department of Labor, Employment and Training Administration (© European Union; reused under the European Commission reuse policy, CC BY 4.0). AI Changing Work has modified this material: match types are shown in AI Changing Work's wording; ESCO occupations are shown with the titles and descriptions of the ESCO records credited under Licenses on this page, not with those in the crosswalk, and these can differ from the ESCO v1.1.0 titles and descriptions the crosswalk was made with; and some matches, including matches to ESCO occupations not among those records, are left out. Where none of that crosswalk's matches for an O*NET occupation is shown, ESCO occupations are shown instead under the ISCO-08 unit groups linked to it by the U.S. Bureau of Labor Statistics crosswalks; these are not matched to it one by one. Neither the European Commission nor the U.S. Department of Labor has approved, endorsed or verified these modifications.

Any AI exposure figures on these occupation pages are published by third parties, not by AI Changing Work, and none is part of the O*NET information. OpenAI publishes task-level scores (MIT License) for O*NET 27.2 task statements; in task tables each is shown next to the 27.2 statement that was scored, and on occupation pages without a task table each is shown next to the O*NET 31.0 task statement with the same task ID, whose wording can differ from the 27.2 statement. OpenAI also publishes occupation-level scores for O*NET-SOC codes in the same release, and any such score is shown on the O*NET occupation with the same code. Anthropic publishes an observed exposure index in the Anthropic Economic Index (CC-BY), and the U.S. Bureau of Labor Statistics publishes relative AI exposure categories (public domain); both are published per SOC code, and each value is shown on every O*NET occupation with that code. In task tables, Anthropic's usage shares, which Anthropic publishes in each Anthropic Economic Index release for O*NET task statements rather than for occupations, and OpenAI's task-level scores are each shown next to the task statement as worded in the O*NET version the source used. On pages with a task table, the O*NET 31.0 task statements are also listed below it, without exposure figures. AI Changing Work matches Anthropic's statements to O*NET task statements by their text, and shows task statements of O*NET-SOC 2010 occupations on the occupations that the O*NET-SOC 2010 to 2019 crosswalk of USDOL/ETA links them to, applying that crosswalk as published, without case-by-case selection; because the same task statement can belong to several O*NET occupations and the crosswalk links some occupations to several others, the same task statement and value can appear on more than one occupation. The International Labour Organization publishes a generative AI exposure index in ILO Working Paper 140 (CC BY 4.0) for ISCO-08 unit groups; AI Changing Work links those groups to O*NET occupations by applying the U.S. Bureau of Labor Statistics ISCO-08 to 2010 SOC and 2010 SOC to 2018 SOC crosswalks as published, without case-by-case selection, and these crosswalks match many groups only in part. Where several unit groups are linked, each group's published value is listed, and any summary shows only the lowest and highest of those values with the number of groups. No published exposure figure is averaged or altered; the shares (%) in the task table are AI Changing Work's calculation, each dividing a published value by the total of the published values for the same occupation in the same column of the same tab. Any employment figures are published by the U.S. Bureau of Labor Statistics for the SOC group containing each occupation. Each source is credited where its figures are shown.

The summaries of recent changes on these occupation pages are not covered by the paragraph above. Each summary was written by AI Changing Work from the source linked with it, and any figures in it are given as AI Changing Work summarised them; they can differ from the source's own wording and from figures shown elsewhere on the same page, so check the source before relying on them. AI Changing Work chose which summaries to show on each occupation by its own judgment, case by case, and the relation labels on them are AI Changing Work's judgment of how closely each summary concerns the occupation, based on the summary's text; a label does not mean that the source names that occupation.

KSCO 코드·명칭: 한국표준직업분류(제8차 개정) — 통계청 고시 제2024-328호 (2024-07-01 고시, 2025-01-01 시행). 저작권법 제7조 제2호의 고시 항목이다. 명칭 표기(가운뎃점·띄어쓰기)는 해설서 2차 정오 반영판의 표기를 따랐으며, 고시 항목표와는 18개 명칭에서 가운뎃점 글리프나 띄어쓰기만 다르다. 통계청은 2025년 10월 국가데이터처로 개편되었다. 직업 페이지의 KSCO 연결은 통계청·국가데이터처의 공식 연계표가 아니다.

Source: France Travail — Répertoire Opérationnel des Métiers et des Emplois (ROME 4.0), version 61, last updated 15 June 2026 (https://www.data.gouv.fr/datasets/repertoire-operationnel-des-metiers-et-des-emplois-rome). Licence Ouverte 2.0.

ROME codes and French titles are reproduced unchanged. The English and Korean titles of the occupations with ACW-FR codes are AI Changing Work's translations of their French ROME titles; these occupations, their AI Changing Work codes and any occupations listed as nearest to them are AI Changing Work's own judgment and are not a France Travail classification or correspondence.

Robot exposure index

Robot exposure figures on the occupation pages (/jobs) are from Russell Legate-Yang and Maxim Massenkoff, "What work can robots do?" (Anthropic, 30 September 2026, https://www.anthropic.com/research/what-work-can-robots-do), robot_exposure data release in the Anthropic EconomicIndex dataset (https://huggingface.co/datasets/Anthropic/EconomicIndex, revision 4e5f69fa), used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). Anthropic produced the task ratings and time shares in this release with Claude Opus 5 using web search. This release includes information from the O*NET 29.3 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Anthropic has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications. AI Changing Work shows each published occupation index on the O*NET occupation with the same O*NET-SOC 2019 code, rounded to two decimal places; the names and descriptions of the exposure tiers are shown in the page's language, and any in languages other than English are AI Changing Work's translations. AI Changing Work does not calculate, combine or adjust these figures. Anthropic does not endorse or sponsor this site or its analyses.

ESCO–O*NET Direct Links v1

The links and match types in this dataset are taken from the ESCO-O*NET Crosswalk, version 1 (September 2022), which pairs O*NET-SOC 2019 with ESCO v1.1.0, by the European Commission (ESCO Secretariat) and the U.S. Department of Labor, Employment and Training Administration (https://esco.ec.europa.eu/en/use-esco/other-crosswalks). © European Union. Reused under the European Commission reuse policy (Commission Decision 2011/833/EU), licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

The ESCO occupation titles are the English preferred labels of the ESCO classification of the European Commission as returned by the ESCO API (https://esco.ec.europa.eu/en/use-esco/use-esco-services-api), retrieved 2026-08-08. © European Union. Reused under the European Commission reuse policy, licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

AI Changing Work has modified this material: (1) only links between O*NET-SOC occupations and ESCO occupations are included; the crosswalk's links to ISCO-08 unit groups are left out, and so are its links to ESCO occupations that are not among the ESCO records AI Changing Work retrieved; (2) ESCO occupations are shown with the titles described above, not with the titles in the crosswalk, and these can differ from the ESCO v1.1.0 titles the crosswalk was made with; (3) where an ESCO occupation's title or description differs from the crosswalk's text, AI Changing Work judged whether the difference changes the occupation's scope, and the links it judged to be changed, or on which the judgments disagreed, are left out; these judgments were made by AI models; (4) match types are given in AI Changing Work's wording in the column relation, next to the crosswalk's own match type in the column crosswalk_match_type; (5) the column review_basis, the O*NET-SOC titles and the order of the rows are AI Changing Work's additions. Neither the European Commission nor the U.S. Department of Labor has approved, endorsed or verified these modifications.

This dataset includes information from the O*NET® 31.0 Database (https://www.onetcenter.org/database.html) by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified all or some of this information: O*NET-SOC codes and titles are reproduced unchanged and are paired with ESCO occupations as in the ESCO-O*NET Crosswalk, with the changes described above. USDOL/ETA has not approved, endorsed, or tested these modifications.

France Occupation Direct Crosswalk v1

This dataset includes information from the O*NET® 31.0 Database (https://www.onetcenter.org/database.html) by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/). O*NET® is a trademark of USDOL/ETA. AI Changing Work has modified all or some of this information: O*NET-SOC codes and titles are reproduced unchanged and paired with ROME 4.0 job profiles by AI Changing Work's judgment; relation labels, statuses and the ACW-FR-* nodes are AI Changing Work's additions. USDOL/ETA has not approved, endorsed, or tested these modifications.

Source: France Travail — Répertoire Opérationnel des Métiers et des Emplois (ROME 4.0), version 61, last updated 15 June 2026 (https://www.data.gouv.fr/datasets/repertoire-operationnel-des-metiers-et-des-emplois-rome). Licence Ouverte 2.0.

ROME codes and titles are reproduced unchanged; the links to O*NET-SOC occupations are AI Changing Work's own judgment and are not a France Travail correspondence. The English and Korean titles of the ACW-FR-* nodes are AI Changing Work's translations of their French ROME titles; these nodes, their AI Changing Work codes and the occupations listed as nearest to them are AI Changing Work's own judgment and are not a France Travail classification or correspondence.