Sources and documentation
Consult the studies we use, identify the edition of each reference and download the data and protocols of our projects.
Selected references, own documents and supporting files. It is not an exhaustive systematic review or independent certification. The conference date records the editorial review of the reference and its identification in this library. It does not indicate new empirical collection or updating of the original study.
19 references in the selection
Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2025-05-20Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: Continuous update pageReference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: Continuous update pageReference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2024Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2025-05-20Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2025Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2025-07-10Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2026-02-24Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2020Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2017-05-25Reference used in the dossier on AI, work and rent. The design and population of each study limit the comparison between their results.
Publication: 2020-05-06Analyzes how digital infrastructure and differences in tasks change the interpretation of AI exposure across countries.
Publication: 2026-03-17Explorer source: 427 occupations and 3,265 tasks. selection identified by review and hash, licensed under CC BY 4.0.
Publication: 2025Public model and price metadata in USD. The effective cost of an execution depends on consumption and the provider used.
Publication: Continuous update pageDocuments provider selection, fallback blocking, and price limits used in the technical protocol.
Publication: Continuous update pageReview of external evidence, with interactive figures, FAQ and income simulation. It does not present its own experiment with companies.
Publication: 2026-10-15Search, comparison and export. Maintains the divergence between the report annex and the tasks of the identified occupation 7126.
Publication: 2026-10-15288 real calls in 24 synthetic tasks: 285 approvals by the JSON criterion. Three format rejections, with correct calculations in subsequent review. Costs reported by API; external individual conciliation pending.
Publication: 2026-10-15Registration open. It involves measuring time, quality and cost per approved delivery. There are no participant results yet.
Publication: 2026-10-15Each base has its version, origin and limitations. The code only uses public data or synthetic tasks.
In 15/10/2026, 100 of 100 questions retrieved the expected font in the limited context used by the product. There were 70 questions about the platform and 30 about Research, without external inference calls.
This assessment measures font retrieval. It does not measure the correctness of generated responses, execution of integrations or security of all flows. The cases were written by the team themselves and used to correct the search; do not constitute an independent validation sample.
The review corrected the excessive influence of common words in search and an old guidance about no logins. Questions, expected references, and retrieved sources are in the report for inspection.
Research uses USD in the new scenarios, costs and collections. The v2 collection protocol keeps previous records in BRL separate. Full article, graphics, tools and navigation have been located. No new inference calls were made.
Public pricing of OpenRouter separate from synthetic test and proposed study with companies. Method prioritizes existing sources and own tests for specific questions. This review did not perform new inference calls.
288 call results, complete responses, filters, exportable figure and execution files. The 1.0.1 protocol registered parameters that were unavailable before calls. The original notes were preserved after reviewing the three format rejections.
Inclusion of ILO WP 166, model catalogue, technical protocol and ISCO-08 explorer with source difference identification in code 7126.
Text revised for clarity. Registration remains open; no participant results were added.
Found an error? Enter the document, version and excerpt viaRovemark contact. Accepted corrections will be described in this record.