An AI-written article can read well and still be wrong. This ten-step check works on any article. It rests on SIFT, the four moves Mike Caulfield published in 2019: stop, investigate the source, find better coverage, and trace claims, quotes and media to the original context.
The ten steps
- Stop. Read the article once before you share or reuse it.
- List the claims. Write down every name, number, quote, date and link you would repeat.
- Investigate the source. Look up the publisher on other sites. A library guide advises against relying on the site’s own “About Us” page.
- Open every link. A dead or missing link is a finding.
- Compare link and claim. Ask whether the page says the same thing, in the same context.
- Check numbers. Find the original figure, who published it, and what it is compared to.
- Trace quotes. Search the exact words and confirm the original wording and setting.
- Check dates. Note when the source was published and whether the claim still holds today.
- Find better coverage. Look for trusted reporting on the same claim, including conflicting accounts.
- Handle paywalls. See the next section.
When a source is paywalled
Look for an open copy first. Bentley University’s library describes Unpaywall as a free browser extension that searches for an open version of an article. If you find none, you have not checked that claim. Mark it unverified or cut it. A snippet or someone else’s summary is not the source.
What to do with the results
Keep the claims that passed. Fix the ones that failed. Remove the ones you could not check.
The Campfire
No commentsNobody has pulled up a log by this one yet. Be the first to say what you make of it.
Held for the desk. It appears after a look.