What is the purpose of newspaper headlines, and who should write them, people or computers?
At the Independent, we work on headlines at both ends of the editorial process. We ask reporters to write them for the first drafts of their articles, because we believe it helps them think about the purpose of the piece. Having to distill a thousand words into one short sentence that accurately conveys the most important element of the story is an excellent way to focus the mind.
Those first draft headlines don’t usually appear in the paper. The final versions that readers see are often written very late at night, after the story has gone through several revisions and fact-checks, and editors have thought about the headline’s nuances and tweaked it to fit the space available.
One of the topics that came up in our recent conversations with the publishers and editors of other weekly newspapers (see “Town by Town,” March 13) was the ways AI is changing local journalism. We get pitches every day from companies promising they can streamline our operation and boost traffic on our website with their AI-generated “content.” And for the tired and lonely editor, headlines are on offer.
For these companies and their clients, the purpose of a headline has little to do with meaning-making. It’s to lure browsers through online searches and get them to click on links. Thus “clickbait” nonsense has come to change the way headlines are written. It used to be a rule in journalism that headlines shouldn’t be questions. Not anymore.
“Why Does Trump Want Greenland? Here’s What to Know,” says the New York Times. “Should you get a measles booster? Here’s what to know,” says the Washington Post. Presumably, these heads perfectly match the questions readers are likely to type into search engines, which is fine, but the superfluous “here’s what to know” drives me nuts. Newspapers tell us things we should know. That goes without saying — at least, it should.
One of our Nantucket colleagues recommended an AI company called Anthropic for writing headlines. I checked it out on a story you might remember from last summer about the dangers of recreational use of erectile dysfunction medications. To be honest, Anthropic’s AI assistant, Claude, was not a bad headline writer. “Recreational ED Drug Use Rises Among Young Adults, Experts Warn,” he suggested — not that different from our own first tries.
A late-night conversation about the real story led us to this instead: “When Recreation Wrecks Your Erection.” I don’t see AI writing headlines like “Bouncing Through the Filth” (on our review of John Waters’s novel Liarmouth) or “Nun’s Farts Are Dunkers to Sigh For” (on our recipe for the Mardi Gras donuts called pets de nonne).
But even if AI could be funny, then what? I’m not worried about my job becoming obsolete. I’m worried about losing something subtler: the chance to solve a late-night puzzle, to have a quick debate with colleagues, to find a shard of poetry or sass for a hidden back and forth between writer and reader.
