Facebook Algorithm Controls: Take Back Your Feed

Facebook’s New Plain-Language Algorithm Controls for Feed and Reels

Facebook is rolling out a new way for users to shape recommendations in Feed and Reels by describing, in plain language, what they want to see more or less of. The update uses large language models to interpret those requests and adjust recommendations in real time, making the algorithm feel more interactive and user-driven.

What Facebook Is Changing

Meta says the new experience lets people directly add and manage the topics they want to see in Feed and Reels through a feature called “Your algorithm.” Users can also type their preferences in their own words, and the system updates recommendations immediately within the same session.

• Users can tell Facebook what they want more of, such as hobbies, interests, or specific content themes.
• Users can also say what they want less of, helping reduce unwanted recommendations.
• The controls apply to both Feed and Reels.
• Changes update in real time rather than waiting for long-term behavior signals to shift.

How the New LLM-Powered System Works

The key shift is that Facebook is using large language models to understand the meaning behind natural-language prompts, including the user’s interests, context, and intent. That means people no longer have to rely only on likes, comments, or watch behavior to influence what appears in their recommendations.

Instead, users can describe what they want in everyday language, and the system translates that into ranking signals. Meta says this makes the algorithm something users can actively talk to rather than a black box that only passively learns from engagement.

Why This Matters for Feed and Reels

This update gives users more direct control over what shows up in the app, which is especially relevant as people look for more useful, relevant, and less repetitive content. It also suggests a broader move toward recommendation systems that respond faster and more transparently to user intent.

• More control over content personalization.
• Faster feedback loops between user input and recommendation changes.
• Better alignment between what users want and what the platform shows.
• A more conversational approach to algorithm tuning.

What Users Can Expect

According to Meta’s description, the new tools are designed to make it easier to refine recommendations without digging through complex settings. The feature appears to be rolling out as part of Facebook’s broader effort to make recommendations more responsive and easier to customize in the moment.

For users who want a more tailored feed, the biggest benefit is simple, direct input. Instead of trying to teach the algorithm indirectly, people can now state their preferences clearly and see the results sooner.

Conclusion

Facebook’s new plain-language algorithm controls mark a major shift in how users shape their Feed and Reels. By combining user prompts with LLM-powered interpretation, Meta is making recommendation tuning more immediate, more intuitive, and more aligned with what people actually want to see.