Across commercial machine-learning services tracked from 2020 to 2022, cases that every system got wrong were more common than independence would predict, and on average only 10% of one system’s improvement fell on cases all the other systems misclassified, though such cases made up 27% of the room for improvement; in a dermatology case, human dermatologists’ outcomes were more homogeneous than the models’.

SocioLens project,

Claim LD09-SL-C23 (v1.0)

Provenance and status

Claim type: Estimate
Publication status: Eligible for publication, with a qualifier
Verification state: Checked against the primary source. Who checked this claim is not recorded for it; claim checks so far were AI-led, and no person has checked claims against their sources.
Does not license: Findings about generative language models
Limitations: Classification systems, not generative models; the excess depends on the independence baseline (see LD09-SL-C26); two human raters; read in the arXiv version 2 of the NeurIPS 2023 paper; the proceedings version was not read
Source (as cited): Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862
Source type: Peer-reviewed conference paper (recorded as “Peer-reviewed conference paper”)
Institutional position: Independent academic (Stanford University-led)
Maturity of the evidence: Emerging evidence
Recorded date: 15 October 2026
Currentness / correction: This is the current version. No correction or concern is recorded on its sources.

Sources this version cites

Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862 (Peer-reviewed conference paper)
Read the original

Version

This is the current version, v1.0. Earlier versions stay available, and their records never change.

Versions: v1.0 · current

Raw record: every recorded field, with its coded values

All 25 canonical Claim fields, verbatim, with the coded values derived from them. The plain-language labels above are read from these.

Claim type (coded): ESTIMATE (as printed: “ESTIMATE”)
Publication status (coded): ELIGIBLE (QUALIFIED) (as printed: “ELIGIBLE WITH QUALIFIER”)
Verification state (coded): CONTENT_VERIFIED_PRIMARY
Source type (coded): PEER_REVIEWED_CONFERENCE_PAPER
Maturity (coded): EMERGING

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD09-SL-C23	—
2	PUBLIC-FACING CLAIM	Across commercial machine-learning services tracked from 2020 to 2022, cases that every system got wrong were more common than independence would predict, and on average only 10% of one system’s improvement fell on cases all the other systems misclassified, though such cases made up 27% of the room for improvement; in a dermatology case, human dermatologists’ outcomes were more homogeneous than the models’.	—
3	CLAIM TYPE	ESTIMATE	base_types: ESTIMATE
4	RESEARCH STRAND	C — Systems, decisions and shared knowledge	—
5	MATURITY	EMERGING	states: EMERGING
6	SOURCE	Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862	source_ids: SRC-SL-TOUPS-ET-AL-NEURIPS-2023
7	SOURCE TYPE	Peer-reviewed conference paper	classes: PEER_REVIEWED_CONFERENCE_PAPER
8	INSTITUTIONAL POSITION	Independent academic (Stanford University-led)	—
9	SCOPE	Commercial classification services (sentiment, speech and images) and a dermatology dataset	—
10	DESIGN	Ecosystem-level comparison of observed shared failure with an independence baseline	—
11	SAMPLE / DATASET	11 datasets and 9 commercial services over three years; 656 dermatology images rated by models and two dermatologists	—
12	MODEL / TOOL / VERSION	Commercial machine-learning APIs of several vendors, and research models in the dermatology case	—
13	COMPARATOR	An independence baseline; human dermatologists	—
14	MAIN RESULT	Systemic failure above baseline in every dataset; 10% of a system’s improvements on systemic failures against 27% of improvable cases; human outcomes more homogeneous than model outcomes in dermatology, though models were more homogeneous on darker skin	—
15	INFERENTIAL TARGET	Whether individual system improvements reach the people every system fails	—
16	DOES NOT LICENSE	Findings about generative language models	—
17	VERIFICATION STATE	CONTENT VERIFIED	state: CONTENT_VERIFIED_PRIMARY · verified_by: UNKNOWN
18	LIMITATIONS	Classification systems, not generative models; the excess depends on the independence baseline (see LD09-SL-C26); two human raters; read in the arXiv version 2 of the NeurIPS 2023 paper; the proceedings version was not read	—
19	CONTRADICTORY / SUPPORTING SOURCES	Baseline dependence: LD09-SL-C26	—
20	GEOGRAPHY / LANGUAGE	Not checked / English	—
21	DISTRIBUTIONAL LEDGER	Models were more homogeneous on images of darker skin, a disparity humans did not show	—
22	R01 / FIELD-MAP LINKS	LD-09, D3	—
23	WHAT WOULD CHANGE THIS CLAIM?	Analyses with difficulty-adjusted baselines, and studies of generative systems	—
24	PUBLICATION STATUS	ELIGIBLE WITH QUALIFIER	state: ELIGIBLE · qualifiers: QUALIFIED
25	RECORDED DATE	2026-10-15	—

Field 22 references resolved against Field Map v0.2: LD-09 (LIVING_DEBATE); D3 (DOMAIN)

The sources as recorded, with their record IDs and coded source types:

Pinned Sources (field 6, version-specific)

“Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862” → Toups et al. (2023), “Ecosystem-level Analysis of…” (SRC-SL-TOUPS-ET-AL-NEURIPS-2023--v1.0) — Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862; class PEER_REVIEWED_CONFERENCE_PAPER (explicit public provenance pin)

Cited by Living Debate versions

Individual gain, collective narrowing? Does AI make science, creative work and shared knowledge more alike? (v0.1) LD-09

Cite this page

SocioLens project. (2026, October 15). Across commercial machine-learning services tracked from 2020 to 2022, cases that every system got wrong were more common than independence would predict, and on average only 10% of one system’s improvement fell on cases all the other systems misclassified, though such cases made up 27% of the room for improvement; in a dermatology case, human dermatologists’ outcomes were more homogeneous than the models’ [Claim LD09-SL-C23, version v1.0]. SocioLens. https://sociolens-web.pages.dev/claims/ld09-sl-c23/v1.0/

@misc{sociolens_ld09_sl_c23_v1_0,
  author = {{SocioLens project}},
  title = {Across commercial machine-learning services tracked from 2020 to 2022, cases that every system got wrong were more common than independence would predict, and on average only 10\% of one system’s improvement fell on cases all the other systems misclassified, though such cases made up 27\% of the room for improvement; in a dermatology case, human dermatologists’ outcomes were more homogeneous than the models’.},
  year = {2026},
  date = {2026-10-15},
  howpublished = {\url{https://sociolens-web.pages.dev/claims/ld09-sl-c23/v1.0/}},
  note = {Claim LD09-SL-C23, version v1.0}
}

The link points to this version, whose record never changes. For a finding, cite the original study as well: SocioLens summarizes research and does not replace it.

Research-of-record provenance

Materialized by sociolens-materializer 0.6.0 at 2026-09-28T20:00:00.000Z; normalization contract SocioLens — Claim Value Vocabulary v1.0.2 — Pre-Spike Materialization Contract (v1.0.2). Content seal SHA-256 5a2da8de49773cd0cd032703103974c472088934f4fe4017489a577ad100114d.

Repository-authored inputs (not Drive exports)

SocioLens claim — LD09-SL-C23 v1.0 (v1.0) — written in the SocioLens repository by Project Lead (Claude, AI) on 2026-10-15; no Drive identity exists. Authority: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and the Owner's choice in the SocioLens project chat on 15 October 2026 to build the LD-09 debate; a new Claim, first version, written from the project's LD-09 source notes, checked against full texts or official abstracts. Block “ENTRY 23 — LD09-SL-C23” SHA-256 e460f1737edef2aafae6b8a9398b1ccf9442c7ca46de4b7d3dcd83e40fbe04ab; fixture fixtures/authored/claims/LD09-SL-C23--v1.0.

Builder decisions (not Drive data)

VERSION = v1.0 — basis: Claim version taken from the repository-authored claim fixture version (SocioLens claim — LD09-SL-C23 v1.0); the first version of a new Claim of the LD-09 debate
CURRENT_POINTER = true — basis: repository-authored claim "SocioLens claim — LD09-SL-C23 v1.0" (Project Lead (Claude, AI), 2026-10-15); basis: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and the Owner's choice in the SocioLens project chat on 15 October 2026 to build the LD-09 debate; a new Claim, first version, written from the project's LD-09 source notes, checked against full texts or official abstracts
FIELD_6_SOURCE_REFERENCE = SRC-SL-TOUPS-ET-AL-NEURIPS-2023--v1.0 — basis: field 6 of ENTRY 23 — LD09-SL-C23 cites "Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862"; the only field-6 segment of this ClaimVersion's repository-authored ledger block.
SOURCE_PROVENANCE_PIN = SRC-SL-TOUPS-ET-AL-NEURIPS-2023--v1.0 — basis: explicit version-pinned public provenance of this ClaimVersion (Amendment v1.2 §6): the pinned field-6 Source record "Toups, Bommasani, Creel, Bana, Jurafsky & Liang, “Ecosystem-level Analysis of Deployed Machine Learning Reveals Homogeneous Outcomes,” NeurIPS 2023, arXiv:2307.05862".