Research opportunity · Who can know about AI? Epistemic dependence and jurisdiction over the field
Whose conversations do data donations capture?
Question
How do people who donate their chatbot conversations to researchers differ from the developers' user bases, and how much do findings from donated data change once donors are weighted to the user population?
Wrong or missing validation targetOne PhD chapter or paperNeeds a budget (medium or higher)Partly addressed
Why it matters
Data donation is one of the few ways researchers study chatbot use without the developer's permission, alongside public logs, and a possible independent check on vendor statistics. But donors are volunteers, and no benchmark of who they are against the real user base was found. If donated data cannot be validated against the population, independent research swaps one unknown sample for another.
The debate lists this among what we still do not know: Are donated or public conversation datasets representative enough to check vendor usage statistics?
Evidence that motivates it
- Researchers studied ChatGPT health advice without the developer’s involvement by collecting 202,590 donated conversations from 1,252 users in India, Nigeria, Brazil and Pakistan (December 2022 to February 2026); an AI judge applying a published rubric rated roughly one in five health conversations high-risk.
Donated conversations from 1,252 users in four countries, collected without the developer. - Analysing 37,845 Australian conversations from WildChat, a public dataset of ChatGPT logs collected outside OpenAI, university researchers classed 61.1% as work-related against 37.3% in OpenAI’s published global figures; to classify them they adopted the classifier prompts and categories of OpenAI-co-authored work.
A public log dataset gives a work share very different from the developer's global figure; the two are not comparable.
A possible design
Recruit donors through a probability-based panel in at least two countries, record demographics and usage frequency, and collect donated histories through data-access rights. Compare donors with non-donors in the same panel and with any published user-base statistics, then reweight key measures (topic shares, work share, sensitive-domain share) and report how far they move.
Data needed
A probability-based survey panel; data-donation tooling under data-access rights; ethics approval; published user-base statistics.
Feasibility
One PhD chapter or paper. Medium to high; panel costs and secure handling of donated data. Developers publish few user-base demographics, so the comparison with the platform population may rest on survey estimates of who uses chatbots.
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).
Donation studies and public datasets exist, but no study was found that measures donor selection against a probability sample and reweights the findings.
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; the data-donation methods literature on social media was 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
- Donated ChatGPT health conversations in four countries (LD08-SL-C23)
Full text read for this debate. Donations from 1,252 users; no comparison with the user base. - Karnam, Dash, Gummadi, Mukherjee, Weber & Zannettou, "Bowling with ChatGPT: On the Evolving User Interactions with Conversational AI Systems," arXiv:2602.01114 (2026) Judged from an abstract, summary or search listing only
Abstract read. 825,000 interactions donated by 300 users through data-access rights; no representativeness benchmark is described in the abstract.
Search queries used
data donation chatbot conversations GDPR representativeness donor selection bias ChatGPT historiesdonated ChatGPT conversations study users data access rights research 2026arXiv id_list: 2602.01114, 2610.18465
Provenance
RO-LD08-06 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.