urn:uuid:92545931-4034-5118-8160-d903484b7dd7 SocioLens — LD-07: AI in the research apparatus New versions of this debate, revisions of its claims, Radar updates and research opportunities. 2026-10-21T00:00:00Z SocioLens urn:uuid:7c7b070b-ecca-5b10-8552-b26d4beb8e3b Claim LD07-HQ-C05 revised to v1.1 2026-10-21T00:00:00Z What changed: The correlation is no longer held back. The main result now gives it: across 469 effects from 70 experiments, GPT-4-derived predictions correlated with the original effect sizes at r = 0.85 (95% CI 0.81–0.88), and at r = 0.92 (95% CI 0.89–0.94) after the authors’ adjustment for sampling error in the original estimates. The limitations now say that these figures come from the authors’ accepted manuscript, not from the Nature version of record. The public-facing claim, the source, the scope, the verification state and the publication status are unchanged; the verification state is the one recorded for version 1.0 and does not cover the two correlations, which were read in the manuscript only. Why: Version 1.0 held back the exact r=.85 figure and its attribution to the Nature version of record (marked HOLD in the R03 LD-07 Canonical Claim Ledger v1.0), because the exact figure had not been extracted line by line from the Nature article. On 21 October 2026 the project owner supplied the authors’ accepted manuscript of the paper, as posted on OSF, which states that it has been accepted for publication in Nature. Its page 4, under "Primary test archive", reports: "In an analysis of 469 effects from 70 experiments, GPT-4-derived predictions were strongly correlated with original effect sizes (r = 0.85, 95% CI = [0.81, 0.88]; r_adj = 0.92, 95% CI = [0.89, 0.94])". The Nature version of record (656:115–122) was still not read, because the publisher blocks the project’s tools, so this version does not claim that the published article prints the same figures; the project owner chose to publish them with that attribution rather than wait for the version of record. The copy read was a print of the manuscript without a text layer, so the figures were read from the page image. The check and this revision were done by AI agents reading the manuscript, not by a person. urn:uuid:944f542b-fb0c-5f02-8a78-8fc784e7dc9a Research opportunity revised to v3: RO-LD07-01 — Do pinned model snapshots keep social-science conclusions stable over time? 2026-10-21T00:00:00Z The note on the nearest work by Chen, Zaharia and Zou no longer says its Harvard Data Science Review issue details are unconfirmed: the DOI's Crossref record (checked 21 October 2026) gives Harvard Data Science Review volume 6, issue 2 (2024), as the References already did. The question, design, literature check and verdict are unchanged. urn:uuid:0ac41e9e-927f-52c1-8102-7443f3eb7cea Debate updated: LD-07 v0.2 — AI in the research apparatus 2026-10-18T00:00:00Z Under what conditions can foundation-model-mediated social research produce valid knowledge? urn:uuid:59231581-cc52-5e9b-b3ca-a7ad58e6c337 New research opportunity: RO-LD07-11 — Does asking in Greek change how AI "silicon samples" answer as Greek respondents? 2026-10-18T00:00:00Z When a language model answers a few European Social Survey Round 11 questions as Greek respondents of a given age, sex, education and region, do its answers come closer to or move further from those of the Greek respondents when the prompt is in Greek rather than in English, overall and across groups? urn:uuid:acf93e2f-ab37-55e8-bb46-3546d850e630 Research opportunity revised to v2: RO-LD07-01 — Do pinned model snapshots keep social-science conclusions stable over time? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:a63c222f-722d-5eaf-840f-bdbd31678fe4 Research opportunity revised to v2: RO-LD07-02 — Does quantization or the serving engine change social-science findings? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The verdict changed from "No study found doing the core design" to "Related work exists" because related work turned up; the question stands. urn:uuid:5b16eefb-0a31-5e10-b0b9-5e5465794778 Research opportunity revised to v2: RO-LD07-03 — How much AI-assisted or agent responding reaches probability panels? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:cbc0b075-b9e5-5bda-a4a9-273caefeed3b Research opportunity revised to v2: RO-LD07-04 — Can synthetic respondents recover strong treatment-effect heterogeneity? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:fa1d4e01-e7ce-5b0e-a9cf-39867aef7af6 Research opportunity revised to v2: RO-LD07-05 — Does post-training change what an LLM measures when it codes text? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:290a26f0-825d-5657-92c2-ddf83a2cc293 Research opportunity revised to v2: RO-LD07-06 — Do AI literature-review tools change the conclusions of social-science reviews? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:e8bb3e23-7ac0-535d-8e38-d3cd180a7766 Research opportunity revised to v2: RO-LD07-07 — How do research ethics boards judge synthetic participants and AI-interpreted data? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The verdict changed from "No study found doing the core design" to "Related work exists" because related work turned up; the question stands. urn:uuid:92b2d69e-d187-5fc5-b1a1-0bff07c85110 Research opportunity revised to v2: RO-LD07-08 — Do silicon samples work outside the North Atlantic, and in which language? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:a5b0bd43-fffb-5a5d-ab97-553797f7f61d Research opportunity revised to v2: RO-LD07-09 — Do LLM judges inflate the quality of AI-written deliberation? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:dc6aeca4-9752-5732-8209-809ee186fbd8 Research opportunity revised to v2: RO-LD07-10 — Does "LLMs can replace coders" depend on which humans you compare with? 2026-10-13T00:00:00Z Literature check added: a second, deeper search for existing work, with its queries, new nearest work and limits. The question and verdict stand. urn:uuid:88909a3f-ab1a-5549-90c5-b3b2a14c3dd3 New research opportunity: RO-LD07-01 — Do pinned model snapshots keep social-science conclusions stable over time? 2026-10-12T00:00:00Z On a fixed battery of published social-science annotation tasks, how often does a substantive conclusion (sign, significance, interval coverage) change between repeated queries of a pinned model snapshot, a moving model alias over 6–12 months, and successive open-weight releases, and can design-based correction absorb those changes? urn:uuid:8c7a03c5-6067-5910-86e7-2ac22f187223 New research opportunity: RO-LD07-02 — Does quantization or the serving engine change social-science findings? 2026-10-12T00:00:00Z With weights, prompts and hardware held fixed, do fp16, 8-bit and 4-bit builds of open-weight models, and different inference engines, change substantive estimates on published annotation and synthetic-respondent tasks more than prompt paraphrasing does? urn:uuid:91206903-9c16-5a42-99e1-1d036f8b840e New research opportunity: RO-LD07-03 — How much AI-assisted or agent responding reaches probability panels? 2026-10-12T00:00:00Z In probability-based panels, what share of completed interviews show AI assistance or autonomous-agent completion when measured with embedded multi-method detection validated against consented chatbot-history linkage, and how does that compare with opt-in platforms measured the same way? urn:uuid:ceadac34-854d-51e6-9af6-2d15ec6956e0 New research opportunity: RO-LD07-04 — Can synthetic respondents recover strong treatment-effect heterogeneity? 2026-10-12T00:00:00Z In probability-sample survey experiments fielded after model training cutoffs and selected for substantial, well-powered moderation, do synthetic samples recover the sign and ranking of conditional treatment effects better than a model that assumes no heterogeneity? urn:uuid:f77e0165-dbc6-58aa-a701-cfb0f28bdfbf New research opportunity: RO-LD07-05 — Does post-training change what an LLM measures when it codes text? 2026-10-12T00:00:00Z Holding base weights fixed, how much do instruction tuning, preference tuning and region-specific fine-tunes shift coding validity and downstream estimates on stance and attitude annotation tasks, and do the shifts track the fine-tune's target culture? urn:uuid:feafa717-1699-5ba7-92a2-b90dd1518293 New research opportunity: RO-LD07-06 — Do AI literature-review tools change the conclusions of social-science reviews? 2026-10-12T00:00:00Z Across a sample of published social-science meta-analyses, do the studies retrieved by AI review tools, compared with those the original review included, change the pooled effect, its heterogeneity or the stated conclusion, and is the missed evidence systematically different (grey literature, non-English studies, null results)? urn:uuid:fda79d01-971a-5703-800d-cdbde4d98843 New research opportunity: RO-LD07-07 — How do research ethics boards judge synthetic participants and AI-interpreted data? 2026-10-12T00:00:00Z How do ethics-board guidance documents across countries classify synthetic participants, LLM coding or interpretation of qualitative data, and transfer of participant data to commercial model APIs, and in a factorial vignette, do board members' decisions and requested safeguards vary with data sensitivity, model hosting and the degree of delegated interpretation? urn:uuid:4be1feeb-ce99-5e3b-b43a-9421db354a8b New research opportunity: RO-LD07-08 — Do silicon samples work outside the North Atlantic, and in which language? 2026-10-12T00:00:00Z Using Afrobarometer, Arab Barometer and Latinobarómetro items fielded after model training cutoffs, does synthetic-sample accuracy at country, subgroup and individual level depend on interview language (for example Swahili or Hausa versus English, Arabic versus English), and do LLM codes of political constructs in low-resource languages pass a replacement test against trained in-country coders? urn:uuid:c5981250-187b-5577-a273-62b6bcfa6c2b New research opportunity: RO-LD07-09 — Do LLM judges inflate the quality of AI-written deliberation? 2026-10-12T00:00:00Z On a corpus mixing human-written and LLM-written deliberative contributions, do LLM judges assign higher construct scores (for example on the Discourse Quality Index or argument quality) to LLM-authored text than expert coders blind to authorship, and is the gap larger when the judge model also wrote the text? urn:uuid:d5e13374-a209-5752-ae49-cd8f151b34bc New research opportunity: RO-LD07-10 — Does "LLMs can replace coders" depend on which humans you compare with? 2026-10-12T00:00:00Z Holding items and codebook fixed, how often does a formal replacement-test verdict flip depending on whether the comparator pool is crowdworkers, trained research assistants, the original study's coders or domain experts, and does the flip rate track construct ambiguity? urn:uuid:e45e89a7-6dcd-5688-9188-341e77865a9f Radar update RADAR-LD07-PEW-SYNTHETIC-SURVEYS-2026 (v1.0) 2026-10-07T00:00:00Z Pew Research Center (30 September 2026) built an AI “digital twin” for each panelist analyzed from three 2026 waves of its American Trends Panel, a nationally representative panel of randomly selected U.S. adults. Each twin was conditioned on that person’s profile and earlier survey answers, and weighted synthetic and human results were compared on nearly 300 questions. With its best-performing configuration (Claude Opus 4.6), synthetic estimates differed from human estimates by an average of 12.4 percentage points. Error was substantial for every subgroup reported, reaching 16.1 points for Republicans and Republican leaners and 15.1 points for Black adults. On 47% of questions at least one answer choice was selected by no synthetic respondent, which happened on no question among humans. Results also differed systematically by AI model, although some toplines matched closely. This strengthens the evidence for LD07-HQ-C06 on selected-average agreement, real-world response variance and subgroup heterogeneity. The report contains no regression or treatment-effect analysis and no individual-level accuracy result for its final configuration, so those parts of the claim are unchanged. The claim wording is unchanged.