Meta rolls out 'Your Algorithm' feed controls on Facebook
The feature lets users add and manage topics they want to see more or less of in their feed and Reels, with recommendations updating in real time.
Meta is introducing a 'Your Algorithm' feature on Facebook that lets users directly adjust their feed recommendations, according to The Verge. The feature is designed to let users add and manage the topics they want to see more or less of in their feed and in Facebook Reels.
Users can select 'interested' or 'not interested' from the three-dot menu on individual posts, and Meta said the buttons may also appear separately below posts. Meta said recommendations react 'immediately' when users input preference signals, and that it may show additional questions asking users more specifically what they liked or disliked.
According to Meta, users can also tell the system in their own words what they want to see more or less of, with recommendations updating in real time. The company said it is using large language models more broadly in its ranking systems, enabling what it describes as an 'algorithm you can actually talk to'.
Meta said recommendations react in real time and adapt to what currently matters, even within the same session.
Earlier versions on Instagram and Threads
'Your Algorithm' was first introduced on Instagram earlier this year, and a similar 'Dear Algo' feature was launched on Threads. Previously, Facebook's algorithm learned indirectly from individual post reactions and viewing history.
Quick answers
What does Facebook's 'Your Algorithm' feature do?
It lets users add and manage the topics they want to see more or less of in their feed and Facebook Reels, and tell the system in their own words what they want to see.
How do I signal a preference on a Facebook post?
Users can select 'interested' or 'not interested' from the three-dot menu on individual posts; Meta said the buttons may also appear separately below posts.
How quickly do Facebook recommendations change?
Meta said recommendations react 'immediately' when users input preference signals, update in real time, and can adapt even within the same session.