Skip to content
ROLearn Intelligence
Disclosure

Use policy

A named person researches every piece, checks every figure against its source, and is accountable for what is published. Here is where AI tools help along the way.

Last updated

The rule

We use AI tools in researching, drafting and editing. No piece is published on a model's say-so: the author named on it chose what to investigate and which sources to use, checked every number, quotation and citation against the source it came from, and holds editorial responsibility for the text. If that is the only thing you read on this page, it is the part that matters.

Where machines are used

  • Drafting and editing.AI tools help turn the author's research, notes and data into prose, and help restructure and shorten it. The author directs the argument, reviews every draft in full, and edits it before publication.
  • Collection and classification at scale. Sorting thousands of mentions, videos, and games into categories, and matching an activation to the content that discusses it. This is machine work by necessity; a person spot-checks samples of it.
  • Forecasting. Where we publish a projection, it comes from a statistical model whose family, inputs, and error metric are stated in the piece.
  • Transcription of interviews, corrected against the audio by a person before quotation.
  • Code. Analysis scripts and this website itself are written with the assistance of coding tools, and reviewed by a person before they run against real data.

Where machines are not trusted

  • Deciding what a piece concludes. A model may propose wording; the finding is the author's, and so is the responsibility for it.
  • Producing a number that then appears in a piece without a person reproducing it.
  • Supplying a quotation, a source, or a citation. Every source listed is one a person opened and checked against the claim it supports.
  • Producing images that could be mistaken for documentary evidence. Illustrations are labelled as illustrations; screenshots are real screenshots.

Why this policy exists

A research publication trades on being checkable, and readers are entitled to know how a piece was made. A model can produce a fluent sentence that contains a citation that does not exist, and it cannot be held responsible for it afterwards. So we draw the line at accountability: a machine may help do work a person verifies, and may not put a claim into print that a person has not verified.

If we get this wrong

If you find a figure, quotation or citation here that does not match its source, or anything that looks like it reached print without the checks described above, tell us at intelligence@rolearn.dev and we will investigate it publicly under the corrections policy.

Use of this site by AI systems

Quoting our research with attribution and a link is welcome, including in an AI-generated answer. Reproducing whole pieces or datasets for model training is not licensed by default; see the terms. Our robots.txt states which automated agents may crawl the site.