An App Can Rewrite This Clickbait Headline — Here's Why

The Artifact News app just started letting users mark articles as clickbait, but starting Friday, the app will actually be recreating clickbait-y headlines for you in real time with the help of OpenAI's GPT-4 Big Language Model. Will make Will be able to write If enough people flag a story as a clickbait headline, Artifact can start showing AI-rewritten headlines to all users.

It's a fascinating — and rapid — evolution of a feature that debuted just last week for Artwork, the new app from the Instagram co-founders that's like a TikTok for text.

In an interview with The Verge, Artifact co-founder Kevin Systrom says the app has already seen "tons" of people using clickbait markers. "It's clear that people are obsessed with ridding the world of clickbait," he says. "It touches a nerve. What if we actually just built a product that let you take action based on the problem you see?"

Clickbait exists partly because in the very competitive environment on social media you are exposed to a lot of news. In order to grab your attention, titles can be overly intriguing or omit information to entice you to click. But on a news-focused app like Artifact, people are already there to read the news, which makes vague headlines feel especially inappropriate.

Clickbait headlines aren't only annoying to Artifact users, though: they can wreak havoc for Artifact's recommendation system. Sometimes, says Systrom, clickbait headlines can tell the system "that you're interested in something you're not really interested in because you clicked to find some important information that was left out of the headline." Was." Are.

Systrom showed me a demo of the AI-powered rewrite process. When a user flags something as a "clickbait title" (you can find the option by long-pressing the article in your feed), they'll see a little loading animation in the place where the offending title used to be, And then the new title will appear as Viewed. Next to the title, there is an asterisk to indicate that it is not the original title. Artifact itself is not capable of rewriting headlines in articles; It can only rewrite them from the feed.

You can get an idea of how this works in the screenshot below provided by Artifact.

Systrom says that the humans at Artifact will also review a queue of headlines that have been most rewritten and can choose to push a revised headline to all Artifact users. Currently this process is manual, but may be automated in the future.

Systrom is confident in the accuracy of the AI-rewritten headlines. "In our experience in testing, it's basically always right," he says. "I don't think we've really found an instance where it's not perfect. We've found instances where it would rewrite the title and it's not necessarily better, but it's not worse. (Hilarious) , as Artifact was refining the feature) , Systrom told me, that it actually writes some of the clickbait itself because it used to see those kinds of headlines online.)

When talking with Systrom, I wondered if Artifact has ambitions to automatically rewrite every headline using AI — if a clickbait marker is already a hit with users, what if they could use AI? Use AI to rewrite your headlines? Like who do you write for? Luckily, it doesn't look like that's in the cards. "I really appreciate and respect the editorial authority of publishers and authors," Systrom says. "I think it's really only targeted in the most severe cases when it causes problems for users."