Last month, the Artifact news app introduced a feature allowing users to flag articles as clickbait. In a recent update, co-founders Kevin Systrom and Mike Krieger have launched an additional functionality that leverages AI to rewrite clickbait headlines.
According to the app developers, if a user identifies a title as clickbait, the app utilizes a GPT-4 model to generate an alternative headline. Initially, Artifact relied on a manual process where articles were assessed and marked as clickbait based on user reports.
With the latest update, if the Artifact team identifies an article as clickbait, the AI system automatically rewrites the headline for all users. A star icon accompanies the revised headline to indicate the AI intervention.
Furthermore, the startup is actively working on developing technology to detect clickbait articles without relying on manual reporting. Once implemented, the app will autonomously identify and rewrite clickbait headlines.
These features demonstrate the app’s commitment to taking on significant editorial responsibility, encompassing the accurate identification of clickbait articles and employing AI to ensure non-misleading headlines. The company is embracing these measures wholeheartedly.
During an interview with TechCrunch in March, Systrom acknowledged that operating a news app involves making editorial decisions.
“Building the algorithm is, in fact, an editorial process because the choices you make in training your algorithm—the objective function, the data selection, inclusion, and exclusion—all involve editorial judgment. It also pertains to how you prioritize different objectives,” he explained.