By Jason Snell
January 17, 2024 12:39 PM PT
Artifact’s killer feature was rewriting bad headlines
[Content warning: Old Man Yells At Cloud.]
So Artifact, the news app startup from the creators of Instagram, is shutting down. There’s been a lot of analysis about the issue, and I agree with much of it.
There’s been a lot of cheap and bad “news content” on the Web for ages now, but in recent years, it seems to have gotten worse. In certain areas—I notice it with film and TV coverage as well as sports—a single bit of original reporting is breathlessly rewritten by dozens of sites, turning a juicy one-sentence quote into a thousand-word-long analysis with lots of backstory and relevant on-site links, all with subheads that answer questions Google users might ask. Once you see it, you can’t unsee it.
For all its boasting about using AI smarts, Artifact didn’t really cut through the noise. Surely an AI content engine could analyze articles and determine which piece was the original and which pieces added very little to the discussion? I’d hope that, eventually, someone will try to counteract this cottage industry of rewriting other people’s scraps for SEO hits with some intelligence that weights the original story over the word salad copycats. (Paging Google—this is supposed to be your area. And Apple News… what are you doing?)
But my favorite feature of Artifact was not its AI-influenced story browsing and categorization. I found it just as weird and off-putting as most other news aggregation apps that don’t allow me complete control over what I see. No, what I loved about Artifact was that you could take a meaningless clickbait headline and have the app read the story and write a new headline based on its contents.
Back in the day, when I learned how to write headlines (for newspapers!), the goal was to give the reader enough information to decide if they wanted to read that article. The Inverted Pyramid started above the story itself, even with key information in a headline and subhead that would let the reader decide if they wanted to bother reading that story.
But in the era of the web and news aggregators, headlines that give away pertinent information have become a lost art. Whole generations of editors have been trained to write coy headlines that will earn a click, even if the people who are clicking will be immediately disappointed by the truth of the story.
What’s most frustrating to me is that once this method of writing headlines is internalized, it’s very hard to break out of the habit. For instance, I subscribe to multiple news sites where you basically must pay in order to read articles… and yet the headlines are still clickbaity! San Francisco Chronicle, when you write, “This Bay Area city has a horrible secret,” I would like to know if it’s my city and if the secret is a serial killer or the lack of a good delicatessen.
This is where Artifact ruled, because its little AI agent would read that story and write a new headline. And if enough people said a headline was clickbait, all other users of the app would now see the AI headline instead of the original. Brilliant.
So let me put this out there to other creators of news apps, aggregators like Apple News but also RSS readers—this is a feature you should knock off. I would absolutely love to be able to see rewritten headlines for news sources that have failed to be forthcoming with factual headlines. Maybe the future of browsing news is an AI-driven service that writes good headlines so that I can make up my own mind about whether I want to read a story or not!
Anyway, I won’t miss Artifact much—but I will miss the perverse joy I felt everytime I had to ask an AI to improve on the work of a human being who has been miseducated about how headlines are supposed to work.
If you appreciate articles like this one, support us by becoming a Six Colors subscriber. Subscribers get access to an exclusive podcast, members-only stories, and a special community.