Claim LD08-SL-C01 (v1.0)

Among notable machine-learning models in the top quarter by training compute, the share built by academic laboratories fell from about 65% in the early 2010s to about 10% in the early 2020s; across about 1.2 million publications, a 10% rise in compute-related keywords in a topic went with a 1.73% fall in the share of academic-only papers.

Provenance and status

Claim type: ESTIMATE (as printed: “ESTIMATE”)
Publication status: ELIGIBLE (PREPRINT) (as printed: “ELIGIBLE WITH PREPRINT QUALIFIER”)
Verification state: CONTENT_VERIFIED_PRIMARY (PREPRINT) (as printed: “CONTENT VERIFIED PREPRINT”; verified by: UNKNOWN)
Does not license: A claim that academia has stopped studying AI or that compute alone causes the shift
Source (field 6, as printed): Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024
Source type (field 7): PREPRINT (as printed: “Preprint”)
Institutional position: Mixed: independent academic (MIT, NYU, Brown, Cornell) and AI-governance nonprofits (Epoch, Centre for the Governance of AI)
Maturity (claim evidence / publication maturity): EMERGING (as printed: “EMERGING”)
Recorded date: 2026-10-16
Currentness / correction: This is the current version. No correction or concern relation is recorded on its Sources.

Version

Pinned version v1.0 (entry LD08-SL-C01--v1.0) — this page never floats to newer evidence.

This pinned version is also the current version.

Versions:v1.0 · current

Pinned Sources (field 6, version-specific)

“Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024” → SRC-SL-BESIROGLU-ET-AL-ARXIV-2401-02452--v1.0 — Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024; class PREPRINT (explicit public provenance pin)

All 25 canonical Claim fields (raw values and normalized values)

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD08-SL-C01	—
2	PUBLIC-FACING CLAIM	Among notable machine-learning models in the top quarter by training compute, the share built by academic laboratories fell from about 65% in the early 2010s to about 10% in the early 2020s; across about 1.2 million publications, a 10% rise in compute-related keywords in a topic went with a 1.73% fall in the share of academic-only papers.	—
3	CLAIM TYPE	ESTIMATE	base_types: ESTIMATE
4	RESEARCH STRAND	A — Who produces knowledge about AI	—
5	MATURITY	EMERGING	states: EMERGING
6	SOURCE	Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024	source_ids: SRC-SL-BESIROGLU-ET-AL-ARXIV-2401-02452
7	SOURCE TYPE	Preprint	classes: PREPRINT
8	INSTITUTIONAL POSITION	Mixed: independent academic (MIT, NYU, Brown, Cornell) and AI-governance nonprofits (Epoch, Centre for the Governance of AI)	—
9	SCOPE	Machine-learning research, 2012–2023; notable models and publications indexed by the authors	—
10	DESIGN	Bibliometric and model-database analysis by author affiliation; regression of academic share on compute-related keywords by topic	—
11	SAMPLE / DATASET	About 650 notable machine-learning models; about 1.2 million publications in 25 sub-fields; 2,333 NeurIPS 2021 papers for hardware use	—
12	MODEL / TOOL / VERSION	Not applicable (study of who builds and studies models)	—
13	COMPARATOR	Industry and mixed industry–academic teams	—
14	MAIN RESULT	Academic share of top-quartile-compute models about 65% (early 2010s) to about 10% (early 2020s); a 10% rise in compute keywords associated with a 1.73% fall in academic-only share; industry-affiliated NeurIPS 2021 groups used hardware about five times more powerful	—
15	INFERENTIAL TARGET	The shift of compute-intensive model building from academia to industry	—
16	DOES NOT LICENSE	A claim that academia has stopped studying AI or that compute alone causes the shift	—
17	VERIFICATION STATE	CONTENT VERIFIED PREPRINT	state: CONTENT_VERIFIED_PRIMARY · qualifiers: PREPRINT · verified_by: UNKNOWN
18	LIMITATIONS	Read in arXiv version 2 (8 January 2024); no peer-reviewed version was found; shares are read from figures and rounded by the authors; the authors note that their notable-model criteria favour industry models; the 1.73% estimate covers 2012–2021; two author organizations work in AI governance and the paper does not state its funding	—
19	CONTRADICTORY / SUPPORTING SOURCES	Compute divide in conference publishing: LD08-SL-C02; industry presence in research on AI’s social effects is lower: LD08-SL-C11	—
20	GEOGRAPHY / LANGUAGE	Global / English	—
21	DISTRIBUTIONAL LEDGER	Elite and well-resourced institutions keep more access than others	—
22	R01 / FIELD-MAP LINKS	LD-08, D3	—
23	WHAT WOULD CHANGE THIS CLAIM?	A count that includes models built with public compute such as national research resources	—
24	PUBLICATION STATUS	ELIGIBLE WITH PREPRINT QUALIFIER	state: ELIGIBLE · qualifiers: PREPRINT
25	RECORDED DATE	2026-10-16	—

Field 22 references resolved against Field Map v0.2: LD-08 (LIVING_DEBATE); D3 (DOMAIN)

Cited by Living Debate versions

LD-08 v0.1

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 a31898bd6a742c69720e59fff52c0549d8b76adba941c64fc2c3362bba310b56.

Repository-authored inputs (not Drive exports)

SocioLens claim — LD08-SL-C01 v1.0 (v1.0) — written in the SocioLens repository by Project Lead (Claude, AI) on 2026-10-16; 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 16 October 2026 to build the LD-08 debate; a new Claim, first version, written from the project's LD-08 source notes, checked against full texts, official web pages or official abstracts. Block “ENTRY 01 — LD08-SL-C01” SHA-256 cb8b6c4d354cae6e468ce203ca662529e4c101d3fdb0923eeb70c9c737b1a9fd; fixture fixtures/authored/claims/LD08-SL-C01--v1.0.

Builder decisions (not Drive data)

VERSION = v1.0 — basis: Claim version taken from the repository-authored claim fixture version (SocioLens claim — LD08-SL-C01 v1.0); the first version of a new Claim of the LD-08 debate
CURRENT_POINTER = true — basis: repository-authored claim "SocioLens claim — LD08-SL-C01 v1.0" (Project Lead (Claude, AI), 2026-10-16); 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 16 October 2026 to build the LD-08 debate; a new Claim, first version, written from the project's LD-08 source notes, checked against full texts, official web pages or official abstracts
FIELD_6_SOURCE_REFERENCE = SRC-SL-BESIROGLU-ET-AL-ARXIV-2401-02452--v1.0 — basis: field 6 of ENTRY 01 — LD08-SL-C01 cites "Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024"; the only field-6 segment of this ClaimVersion's repository-authored ledger block.
SOURCE_PROVENANCE_PIN = SRC-SL-BESIROGLU-ET-AL-ARXIV-2401-02452--v1.0 — basis: explicit version-pinned public provenance of this ClaimVersion (Amendment v1.2 §6): the pinned field-6 Source record "Besiroglu, Bergerson, Michael, Heim, Luo & Thompson, “The Compute Divide in Machine Learning: A Threat to Academic Contribution and Scrutiny?”, arXiv:2401.02452, 2024".