[DigitalToday reporter Jinju Hong (홍진주)] Facebook is introducing a feature called “Your Algorithm” that lets users directly adjust feed algorithms.

The Verge reported on Oct. 13 that the feature is designed to let users add and manage topics they want to see more or less of in their feed and in Facebook Reels.

The key is that it brings control over the recommendation feed closer to the front. Users can select “Interested” or “Not Interested” from the three-dot menu on individual posts. Facebook may also show these buttons separately below posts. When users enter preference signals, recommendations change immediately. Meta said recommendations “respond instantly.”

The feature goes beyond simple button taps. Meta said it can show additional questions that ask more specifically what users liked or disliked. It also said users can easily tell it, in their own words, what they want to see more or less of, and recommendations are updated in real time. The plan is to improve recommendation accuracy by reflecting not only short reaction choices but also sentence-level feedback.

The change is part of Meta’s expansion of AI-based personalized recommendation tools applied to Facebook. Meta said it is using large language models (LLMs) more broadly in its ranking systems. It highlighted that LLMs can reflect interests, context and user intent contained in natural language. Meta said this approach makes it possible to have an algorithm that can “actually hold a conversation.” It said that when users say what they want to see more or less of, recommendations “respond in real time and adapt to what matters now, even within the same session.”

Meta has already applied similar functions to other services. “Your Algorithm” was introduced first on Instagram earlier this year. A similar feature called “Dear Algo” was also launched on Threads. By extending the same direction of features to Facebook, Meta is continuing a trend of redesigning recommendation systems across its major social platforms around user input.

For users, the key change is that it can reduce some of the opacity of recommendation feeds. Previously, algorithms mainly learned indirectly from reactions to individual posts or viewing history. Now, users can directly include or exclude specific topics, mark preferences at the post level, and, if needed, explain intent in natural language.

As a result, Facebook’s recommendation structure is more likely to shift from a focus on simple engagement metrics toward interpreting user intent. Since Meta has said it will more fully connect LLMs to its recommendation systems, Facebook feeds are expected to be run in a way that reflects users’ explicit feedback more quickly.