How we use AI
AI tools help us research products and edit what we publish. The criteria every ranking is judged on are always reviewed by a person.
Where AI helps
- Collecting each product’s own documentation, website text and release history.
- Finding the exact lines in those sources that support or limit a score.
- Drafting and editing the text of our articles.
What a person always reviews
Before a ranking is published, a person reviews its criteria and how much each one counts. A criterion that isn’t meaningful to someone choosing software, or that can’t be checked against a source, is rewritten before any product is scored on it.
What stays the same either way
- Every score cites an exact quote from the product’s own source, with a link and the date it was checked.
- Each quote is checked word for word against the page it came from before it is published.
- The order of a ranking is computed from the weighted scores. Nobody, and no tool, picks the winner by hand.
- Unless an article says otherwise, scores come from documentation and release data, not from hands-on testing.