Research opportunity · Who can know about AI? Epistemic dependence and jurisdiction over the field
Who writes research on AI's social effects outside arXiv?
Question
Across social-science and humanities journals, books, policy reports and arXiv, what share of research on AI's social effects comes from computer-science teams, social scientists, mixed teams, industry and government, and how has that share changed since 2014?
Not yet measuredOne PhD chapter or paperLow to medium costPartly addressed
Why it matters
The claim that computer science is absorbing the study of AI's social effects (the jurisdiction position, P4) rests mainly on arXiv papers; the main study excludes social-science journals. If journals, books and policy reports tell a different story, the debate's jurisdiction picture changes. The disciplinary map outside arXiv is not measured.
The debate lists this among what we still do not know: Is computer science absorbing research on AI's social effects outside technical preprint servers?
Evidence that motivates it
- In about 102,000 arXiv AI papers (2014–2024), the share of societally oriented sentences written by computer-science-only teams rose from 49.0% to 71.2%, while the share written by teams including social scientists or humanists fell from 25.7% to 3.8%, although such teams wrote the most societally oriented papers (20.9% of sentences, against 7.8% for computer-science-only teams).
The rise of computer-science-only teams is measured on arXiv papers only. - Of the papers published at the FAccT conference from 2018 to 2021, 61.3% were labelled STEM and 29.7% humanities, social sciences and arts; summarising a survey of conference affiliates, the authors list concerns including “out-sized industry connections.”
The disciplinary mix of one conference, 2018 to 2021. - Of 84 AI ethics guidelines collected worldwide by 2019, private companies issued the largest share (19, or 22.6%), ahead of governmental agencies (18) and academic and research institutions (9).
Companies issued the largest share of AI ethics guidelines collected by 2019. - In April 2025 Anthropic set up an Economic Advisory Council of “distinguished economists” to advise it on AI’s economic effects and to “inform the research agenda for the Anthropic Economic Index.”
A developer's council of economists informs the agenda of its usage research.
A possible design
Draw a stratified sample of publications on AI's social effects from journal databases, book catalogues, policy-report repositories and arXiv (2014–2026), code authors' disciplines and sectors from affiliations and CVs, and estimate shares by venue type and year. Use the same coding for arXiv as the existing arXiv study so the two can be compared.
Data needed
Bibliographic records from journal and book databases; policy-report collections; affiliation and CV data; a coding protocol for discipline and sector.
Feasibility
One PhD chapter or paper. Low to medium; database access and coding time. Database coverage favours English-language journals; report language coverage alongside the shares.
Has this been done?
Partly done already: related work covers some of this question (first-pass screen of 16 October 2026; not yet checked in depth).
Bibliometric studies count AI publications in the social sciences, and one study measures team composition on arXiv, but no study was found that compares who writes on AI's social effects across arXiv, journals, books and policy reports.
How strong this evidence is
- Who searched and checked: .
- What was read: 1 of the 2 nearest works below was judged from an abstract, summary or search listing only; it is marked. Not every entry says how much of the work was read.
- Where the search could not reach: Screened by web searches and arXiv record lookups on the date shown; Scopus, Web of Science and book catalogues were not searched. Absence of a hit is not proof that the work does not exist.
- Before building a project on it, run your own search in a scholarly database such as Scopus, Web of Science, Google Scholar or SSRN.
Nearest work found
- arXiv team-composition study (LD08-SL-C12)
Full text read for this debate. Team composition of societally oriented writing in arXiv AI papers only. - Prieto-Gutierrez, Segado-Boj & Da Silva França, "Artificial intelligence in social science: A study based on bibliometrics analysis," Human Technology 19(2) (2023); arXiv:2312.10077 Judged from an abstract, summary or search listing only
Abstract read. Counts more than 19,408 AI-related social-science articles (2013–2022) by sub-area and country; not authors' disciplines or sectors.
Search queries used
disciplinary composition authors research on social impact of AI journals versus arXiv bibliometric study computer science social sciencewho studies AI's societal effects bibliometric analysis author discipline industry share social science journalsarXiv id_list: 2312.10077, 2506.08738
Provenance
RO-LD08-05 v1. Drafted by Claude (SocioLens Project Lead), AI-drafted from the LD-08 debate on 2026-10-16. Not yet reviewed. Not yet reviewed by a domain expert. This is a research hypothesis, not a finding.