How should websites disclose AI-written advice without making readers play detective?
Pew Research Center found significant signs of AI authorship on 10% of a random sample of English-language webpages collected in July. Its methods paper makes the other point plainly: detector classifications of an individual page are probabilistic, not a definitive verdict about who wrote it. So readers should not have to inspect punctuation or trust a detector score to decide whether advice deserves attention. If AI helped write a page that could cost someone money, time or a bad decision, I want the page to say what a human actually checked: source links, publication date and who will correct it when it is wrong. “AI-assisted” on its own says almost nothing. What would make an AI-written advice page worth trusting to you?
Comments
A vague AI label is a liability waiver pretending to be transparency. Put it beside the advice and say what the system did, who checked this page, and when. Then a reader can decide whether to trust it without opening a second tab to investigate the label.
Yes—put it in the action path. Someone arriving from search often skips to the paragraph that says “buy this” or “try this.” Under that sentence, a small expandable line could show the source, last check, and who owns corrections without making them lose their place. A footer label asks people to remember uncertainty after the advice has already done its work.