Meta, the parent company of Instagram and Facebook, has been facing criticism for its lack of transparency in how it recommends content to users and utilizes user data. In an effort to address these concerns, the company has released a comprehensive analysis of its social media algorithms, shedding light on how content is suggested on both platforms.
The analysis consists of 22 “system cards” that provide detailed information about the AI systems used in different sections of Instagram and Facebook, as well as the user data required for their functioning. Each card offers an overview of the AI system, an explanation of its operation, customizable controls for users, and additional insights into content delivery.
For instance, Instagram’s Explore feature, which displays content from accounts users don’t follow, operates through a three-step process. It curates public content adhering to quality guidelines, considers user engagement signals, and ranks content based on predicted user interest. Users can influence this process by saving or indicating disinterest in certain content.
On Facebook, the AI system collects potential posts from friends and followed Pages, taking into account factors such as creator identity, past interactions, and popularity. The algorithm then ranks posts based on perceived relevance and displays them in the user’s feed. Users have the option to customize their feed by unfollowing, indicating preferences, and adjusting content visibility.
To enhance transparency and user control, Instagram and Facebook are introducing features like the “Why Am I Seeing This?” tool and the ability to mark content as “Interested” or “Not Interested.” These additions aim to help users understand content recommendations and tailor their experience.
Overall, Meta’s release of algorithmic insights seeks to address concerns around content recommendation and empower users to have more control over their social media experience.
