In 171,394 papers from 57 computer-science conferences (2000–2019), after deep learning took off in 2012 large technology firms published 44 more papers a year per AI conference than a synthetic counterfactual predicts and elite universities 40 more, while mid-tier and lower-tier universities published fewer; the authors attribute the gap to unequal access to compute.

SocioLens project,

Claim LD08-SL-C02 (v1.0)

Provenance and status

Claim type: Estimate
Publication status: Eligible for publication, with qualifiers: working paper
Verification state: Checked against the primary source; the source is a preprint. 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: A measure of influence or quality, or a statement about research on AI’s social effects
Limitations: Working paper (arXiv version 1); no journal version found; the mid- and lower-tier losses are reported with different sizes in the introduction (14 and 5 fewer) and in the body, so this claim gives no number for them; the authors attribute the gap in part to compute, from a text analysis of papers; placebo tests and alternative rankings give the same pattern; the ranking is US- and QS-centric
Source (as cited): Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020
Source type: Working paper (recorded as “Academic working paper”)
Institutional position: Independent academic (Ivey Business School and Virginia Tech)
Maturity of the evidence: Emerging evidence
Recorded date: 16 October 2026
Currentness / correction: This is the current version. No correction or concern is recorded on its sources.

Sources this version cites

Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020 (Working 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 (WORKING_PAPER) (as printed: “ELIGIBLE WITH WORKING-PAPER QUALIFIER”)
Verification state (coded): CONTENT_VERIFIED_PRIMARY (PREPRINT)
Source type (coded): WORKING_PAPER
Maturity (coded): EMERGING

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD08-SL-C02	—
2	PUBLIC-FACING CLAIM	In 171,394 papers from 57 computer-science conferences (2000–2019), after deep learning took off in 2012 large technology firms published 44 more papers a year per AI conference than a synthetic counterfactual predicts and elite universities 40 more, while mid-tier and lower-tier universities published fewer; the authors attribute the gap to unequal access to compute.	—
3	CLAIM TYPE	ESTIMATE	base_types: ESTIMATE
4	RESEARCH STRAND	A — Who produces knowledge about AI	—
5	MATURITY	EMERGING	states: EMERGING
6	SOURCE	Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020	source_ids: SRC-SL-AHMED-WAHED-ARXIV-2010-15581
7	SOURCE TYPE	Academic working paper	classes: WORKING_PAPER
8	INSTITUTIONAL POSITION	Independent academic (Ivey Business School and Virginia Tech)	—
9	SCOPE	Papers at 57 computer-science conferences, 2000–2019	—
10	DESIGN	Generalized synthetic control treating 2012 as a shock to AI conferences, with non-AI conferences as controls; text analysis of compute mentions	—
11	SAMPLE / DATASET	171,394 papers	—
12	MODEL / TOOL / VERSION	Not applicable	—
13	COMPARATOR	A synthetic counterfactual built from non-AI computer-science conferences	—
14	MAIN RESULT	Fortune 500 Global technology firms +44 papers a year per AI conference against the counterfactual (from an average of 23); QS top-50 universities +40; mid-tier (QS 201–300) and lower-tier (QS 301–500) universities fewer	—
15	INFERENTIAL TARGET	Who gained and lost participation in AI research after deep learning	—
16	DOES NOT LICENSE	A measure of influence or quality, or a statement about research on AI’s social effects	—
17	VERIFICATION STATE	CONTENT VERIFIED PREPRINT	state: CONTENT_VERIFIED_PRIMARY · qualifiers: PREPRINT · verified_by: UNKNOWN
18	LIMITATIONS	Working paper (arXiv version 1); no journal version found; the mid- and lower-tier losses are reported with different sizes in the introduction (14 and 5 fewer) and in the body, so this claim gives no number for them; the authors attribute the gap in part to compute, from a text analysis of papers; placebo tests and alternative rankings give the same pattern; the ranking is US- and QS-centric	—
19	CONTRADICTORY / SUPPORTING SOURCES	LD08-SL-C01	—
20	GEOGRAPHY / LANGUAGE	Global / English	—
21	DISTRIBUTIONAL LEDGER	Mid- and lower-tier universities lose relative participation; historically Black colleges and universities and Hispanic-serving institutions are heavily under-represented throughout; no post-2012 change was found for historically Black colleges and universities, and Hispanic-serving institutions were not tested	—
22	R01 / FIELD-MAP LINKS	LD-08, D3	—
23	WHAT WOULD CHANGE THIS CLAIM?	A replication after 2019 or with public compute provision as a treatment	—
24	PUBLICATION STATUS	ELIGIBLE WITH WORKING-PAPER QUALIFIER	state: ELIGIBLE · qualifiers: WORKING_PAPER
25	RECORDED DATE	2026-10-16	—

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

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

Pinned Sources (field 6, version-specific)

“Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020” → Ahmed & Wahed (2020), “The De-democratization of AI: Deep…” (SRC-SL-AHMED-WAHED-ARXIV-2010-15581--v1.0) — Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020; class WORKING_PAPER (explicit public provenance pin)

Cited by Living Debate versions

Who can know about AI? Epistemic dependence and jurisdiction over the field (v0.1) LD-08

Cite this page

SocioLens project. (2026, October 16). In 171,394 papers from 57 computer-science conferences (2000–2019), after deep learning took off in 2012 large technology firms published 44 more papers a year per AI conference than a synthetic counterfactual predicts and elite universities 40 more, while mid-tier and lower-tier universities published fewer; the authors attribute the gap to unequal access to compute [Claim LD08-SL-C02, version v1.0]. SocioLens. https://sociolens-web.pages.dev/claims/ld08-sl-c02/v1.0/

@misc{sociolens_ld08_sl_c02_v1_0,
  author = {{SocioLens project}},
  title = {In 171,394 papers from 57 computer-science conferences (2000–2019), after deep learning took off in 2012 large technology firms published 44 more papers a year per AI conference than a synthetic counterfactual predicts and elite universities 40 more, while mid-tier and lower-tier universities published fewer; the authors attribute the gap to unequal access to compute.},
  year = {2026},
  date = {2026-10-16},
  howpublished = {\url{https://sociolens-web.pages.dev/claims/ld08-sl-c02/v1.0/}},
  note = {Claim LD08-SL-C02, 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 3c97e209c20fd6446613544bc9a5c0c52904fa17c3b3d01f62bdf0228a963c5f.

Repository-authored inputs (not Drive exports)

SocioLens claim — LD08-SL-C02 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 02 — LD08-SL-C02” SHA-256 597db2911eb94aaa9419f37e1c3d48e2446b580dbeea262ec5e43c32311b8ca8; fixture fixtures/authored/claims/LD08-SL-C02--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-C02 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-C02 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-AHMED-WAHED-ARXIV-2010-15581--v1.0 — basis: field 6 of ENTRY 02 — LD08-SL-C02 cites "Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020"; the only field-6 segment of this ClaimVersion's repository-authored ledger block.
SOURCE_PROVENANCE_PIN = SRC-SL-AHMED-WAHED-ARXIV-2010-15581--v1.0 — basis: explicit version-pinned public provenance of this ClaimVersion (Amendment v1.2 §6): the pinned field-6 Source record "Ahmed & Wahed, “The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research,” arXiv:2010.15581, 2020".