In 11 AI-exposed Danish occupations, Humlum and Vestergaard estimate earnings and recorded-hours effects close to zero two years after ChatGPT: the confidence intervals of their pooled estimates exclude average effects larger than 2% over that horizon.

SocioLens project,

Claim LD02-HQ-C02 (v1.2)

Provenance and status

Claim type: Estimate
Publication status: Eligible for publication, with qualifiers: scope
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: Global aggregate null; no hiring effect; no task or occupational change
Limitations: One study, not independently replicated; Denmark; 11 selected exposed occupations, not the whole economy; self-reported adoption/use; pre-agentic horizon; the bound is on average effects and does not show that the effect is exactly zero
Source (as cited): Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026
Source type: Working paper (recorded as “Academic working paper”)
Institutional position: Independent academic
Maturity of the evidence: Rapidly maturing evidence
Recorded date: 22 October 2026
Currentness / correction: This is the current version. No correction or concern is recorded on its sources.

Sources this version cites

Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026 (Working paper)
Read the original

Version

This page shows version v1.2. Its record never changes, even when newer evidence arrives; the site may add a note to the page or withhold a figure not cleared for public use.

This pinned version is also the current version.

Versions: v1.0, v1.1, v1.2 · current

What changed in this version

This version supersedes version v1.1 (LD02-HQ-C02--v1.1), which stays available with its record unchanged.

Changed fields: 2 PUBLIC-FACING CLAIM; 14 MAIN RESULT; 18 LIMITATIONS; 25 RECORDED DATE
What changed: The claim no longer says “precise null effects” and “ruling out effects larger than 2%”. It now says that the estimated effects are close to zero and that the confidence intervals of the pooled estimates exclude average effects larger than 2% over the two-year horizon, so the bound is not read as proof of no effect at all. The main result names where the paper reports this (Section 3.2; Figure 7, Panel (a)) and adds that the dynamic estimates exclude changes larger than 3% (Figure 5, Panel (b)). The limitations now say that this is one study, not independently replicated, covering 11 occupations rather than the whole economy. The figures, the source and the other fields are unchanged.
Why: An external research-integrity review of the site on 22 October 2026 (an AI reviewer, not a person) found the claim supported but asked that “null” not be read as proof of a literally zero effect, and that the figure be named. SocioLens re-read the March 2026 revision of NBER Working Paper 33777 on 22 October 2026. Its introduction says the estimates for earnings, recorded hours and wages “all center on zero, with confidence intervals ruling out average effects larger than 2%”; Section 3.2 says the dynamic estimates “rule out changes larger than 3%, and our pooled difference-in-differences estimates rule out effects larger than 2%”, and footnote 14 places the pooled estimates in Figure 7, Panel (a). The reviewer also noted that the maturity label “rapidly maturing” should not suggest independent replication or economy-wide coverage; the label is kept and the limitations now say so. The check and this revision were done by an AI agent reading the paper, not by a person.

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 (SCOPE) (as printed: “ELIGIBLE WITH SCOPE QUALIFIER”)
Verification state (coded): CONTENT_VERIFIED_PRIMARY
Source type (coded): WORKING_PAPER
Maturity (coded): RAPIDLY_MATURING

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD02-HQ-C02	—
2	PUBLIC-FACING CLAIM	In 11 AI-exposed Danish occupations, Humlum and Vestergaard estimate earnings and recorded-hours effects close to zero two years after ChatGPT: the confidence intervals of their pooled estimates exclude average effects larger than 2% over that horizon.	—
3	CLAIM TYPE	ESTIMATE	base_types: ESTIMATE
4	RESEARCH STRAND	E — Wages, hours, employment and transitions	—
5	MATURITY	RAPIDLY MATURING	states: RAPIDLY_MATURING
6	SOURCE	Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026	source_ids: SRC-LD02-HUMLUM-VESTERGAARD-NBER-33777
7	SOURCE TYPE	Academic working paper	classes: WORKING_PAPER
8	INSTITUTIONAL POSITION	Independent academic	—
9	SCOPE	Denmark; 11 AI-exposed occupations; roughly two-year post-ChatGPT horizon	—
10	DESIGN	Adoption surveys linked to administrative labour records; difference-in-differences	—
11	SAMPLE / DATASET	Approximately 25,000 workers / 7,000 workplaces linked to Danish administrative records	—
12	MODEL / TOOL / VERSION	ChatGPT and workplace chatbots, including in-house systems	—
13	COMPARATOR	Adopters/users vs relevant non-adopters/non-users under study design	—
14	MAIN RESULT	Pooled difference-in-differences estimates for earnings, recorded hours and wages centre on zero, and their 95% confidence intervals exclude average effects larger than 2% (Section 3.2; Figure 7, Panel (a)); the dynamic estimates exclude changes larger than 3% (Figure 5, Panel (b)); tasks nevertheless reorganise	—
15	INFERENTIAL TARGET	Short-run incumbent earnings/hours effect	—
16	DOES NOT LICENSE	Global aggregate null; no hiring effect; no task or occupational change	—
17	VERIFICATION STATE	CONTENT VERIFIED	state: CONTENT_VERIFIED_PRIMARY · verified_by: UNKNOWN
18	LIMITATIONS	One study, not independently replicated; Denmark; 11 selected exposed occupations, not the whole economy; self-reported adoption/use; pre-agentic horizon; the bound is on average effects and does not show that the effect is exactly zero	—
19	CONTRADICTORY / SUPPORTING SOURCES	Consistent with broad short-run nulls; the same paper also finds no differential change in employment (total hours), job churn or the early-career share at workplaces that encourage chatbot use through December 2024, while Nationalbank 2026 finds lower employment growth at AI-adopting firms by late 2025 with a different adoption measure, unit and horizon	—
20	GEOGRAPHY / LANGUAGE	Denmark / English	—
21	DISTRIBUTIONAL LEDGER	Early-career (22–25) employment fell in several of the occupations, but the early-career share did not change differentially at adopting workplaces (effects above about a third of a percentage point ruled out); adopters move into higher-paying occupations; no separate early-career earnings/hours estimate is reported	—
22	R01 / FIELD-MAP LINKS	D2, L-A, L-B, L-C, R-1	—
23	WHAT WOULD CHANGE THIS CLAIM?	Later waves showing material divergence in earnings/hours	—
24	PUBLICATION STATUS	ELIGIBLE WITH SCOPE QUALIFIER	state: ELIGIBLE · qualifiers: SCOPE
25	RECORDED DATE	2026-10-22	—

Field 22 references resolved against Field Map v0.2: D2 (DOMAIN); L-A (LENS); L-B (LENS); L-C (LENS); R-1 (LENS)

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

Pinned Sources (field 6, version-specific)

“Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026” → Humlum & Vestergaard (2026), “Still Waters, Rapid Currents” (SRC-LD02-HUMLUM-VESTERGAARD-NBER-33777--v1.0) — Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026; class WORKING_PAPER (explicit public provenance pin)

Cited by Living Debate versions

Aggregate null or distributional shift? Employment (v0.3) LD-02

Cite this page

SocioLens project. (2026, October 22). In 11 AI-exposed Danish occupations, Humlum and Vestergaard estimate earnings and recorded-hours effects close to zero two years after ChatGPT: the confidence intervals of their pooled estimates exclude average effects larger than 2% over that horizon [Claim LD02-HQ-C02, version v1.2]. SocioLens. https://sociolens-web.pages.dev/claims/ld02-hq-c02/v1.2/

@misc{sociolens_ld02_hq_c02_v1_2,
  author = {{SocioLens project}},
  title = {In 11 AI-exposed Danish occupations, Humlum and Vestergaard estimate earnings and recorded-hours effects close to zero two years after ChatGPT: the confidence intervals of their pooled estimates exclude average effects larger than 2\% over that horizon.},
  year = {2026},
  date = {2026-10-22},
  howpublished = {\url{https://sociolens-web.pages.dev/claims/ld02-hq-c02/v1.2/}},
  note = {Claim LD02-HQ-C02, version v1.2}
}

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 bd796de6a2c90433027219ec6d72c2fb1aadd60da868963612d5e478201c94d3.

Repository-authored inputs (not Drive exports)

SocioLens claim revision — LD02-HQ-C02 v1.2 (v1.2) — written in the SocioLens repository by Project Lead (Claude, AI) on 2026-10-22; no Drive identity exists. Authority: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and docs/PROJECT.md "Research revisions" (a published claim changes only through a new version that states what changed and why). Block “ENTRY 02 — LD02-HQ-C02” SHA-256 ccc4d138a6468092cfe0b44aed023a07c790f087b6ab1ebb88b275567c5fcb06; fixture fixtures/authored/claim-revisions/LD02-HQ-C02--v1.2.

Builder decisions (not Drive data)

VERSION = v1.2 — basis: Claim version taken from the repository-authored revision fixture version (SocioLens claim revision — LD02-HQ-C02 v1.2); a new version of LD02-HQ-C02 v1.1, which stays routable (AGENTS.md research integrity rule 3)
CURRENT_POINTER = true — basis: repository-authored revision "SocioLens claim revision — LD02-HQ-C02 v1.2" (Project Lead (Claude, AI), 2026-10-22); basis: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and docs/PROJECT.md "Research revisions" (a published claim changes only through a new version that states what changed and why)
FIELD_6_SOURCE_REFERENCE = SRC-LD02-HUMLUM-VESTERGAARD-NBER-33777--v1.0 — basis: field 6 of ENTRY 02 — LD02-HQ-C02 cites "Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026"; the field-6 text of this ClaimVersion's repository-authored ledger block; the same Source record LD02-HQ-C02 v1.1 cites.
SOURCE_PROVENANCE_PIN = SRC-LD02-HUMLUM-VESTERGAARD-NBER-33777--v1.0 — basis: explicit version-pinned public provenance of this ClaimVersion (Amendment v1.2 §6): the pinned field-6 Source record "Humlum & Vestergaard, “Still Waters, Rapid Currents,” NBER Working Paper 33777, revised March 2026".