AIGenerated
EN · IT · ES

Meta Introduces Your Algorithm A Chat-Worthy Dashboard to Control What You See on Facebook

A new personalization hub lets you type or pick topics you want more and less of, and your Feed and Reels respond right away.

Apps and productsMeta Introduces Your Algorithm A Chat-Worthy Dashboard to Control What You See on FacebookHerald of AI

What you could do with it

  • A Facebook user could open the three-dot menu on any post and type what they want to see more or less of, and their Feed and Reels would adjust right away 2.
  • A Facebook user could browse the 'Your algorithm' dashboard to see a summary of recent viewing activity and rearrange their 'Interested In' and 'Not Interested In' lists, getting instant changes across both surfaces 1.
  • A Facebook user could give follow-up feedback in their own words after tapping 'Interested' or 'Not interested' on a post, helping the system refine recommendations within the same session 2.

How the page looks and works

Your algorithm lives behind the three-dot menu on every post and shows a single editable view of the interests shaping both Feed and Reels 2. At the top is a one-sentence Lately Summary that describes recent viewing activity 1. Below are two lists: "Interested In," which comes preloaded with preferences guessed from earlier behavior, and "Not Interested In," which holds subjects people have clearly signed up as uninteresting, or dismissed by swiping them away or hiding them . People can add interests in their own words, move them between lists, or remove them, and every change applies responsively to both surfaces while continuing to update as users give feedback on individual posts .

Turning natural language into ranked posts

The system maps freeform text onto the ranking funnel from retrieval through reranking to the device 1. For each Interested In entry, four dedicated retrieval sources pull candidates: direct search, clusters of closely related topics, the broader content graph, and item-to-item similarity, then an embedding-based comparison filters by meaning before late-stage promotion gives fresh interests a chance to compete alongside normal content . A control layer spaces recommendations for the same interest over time and caps daily contribution per interest so new topics show up without flooding the feed . For Not Interested In, layered filtering runs at several ranking stages using the same embedding comparison, pushes down near matches after ranking through an LLM reranker, and uses cache invalidation on the device to clear already-queued items . Topic-free requests such as reducing AI-generated or sexually suggestive material map to dedicated demotion signals, while creator-specific blocks filter named accounts directly .

Making changes appear immediately

Even tuned ranking leaves gaps when matching content was fetched earlier or cached on the phone. To close those gaps, a new supplementary tail load fires outside the normal fetch cycle when an Interest is added, pulling fresh candidates right away, while on-device cache invalidation removes pre-ranked items that conflict with updates 1. The whole pipeline reads from a single shared store so the lists shown on the page are exactly what power ranking for both Feed and Reels . Facebook says the rollout is gradual, and follow-up prompts may appear after users tap Interested or Not interested to collect richer natural-language feedback in real time 2.

This story reports on a trend for information only. It is not advice or encouragement to do any of it.

Key facts

  • Your algorithm adds a three-part preference page above the regular Feed.
  • Each Interested In topic drives four retrieval sources instead of relying only on engagement signals.
  • Not Interested In entries are enforced at multiple ranking stages plus on-device cache clearing.
  • A supplementary tail load fetches fresh candidates immediately after preference changes.
  • The feature draws from both freeform natural-language input and existing engagement history.

Sources

  1. Your Algorithm: Giving People More Ways to Personalize Facebook Recommendations — engineering.fb.com
  2. Your Algorithm, Your Words: New Ways to Shape What You See on Facebook — about.fb.com

If this story concerns you or your company, or uses something that belongs to you, ask for a correction or for it to be taken down. How requests work

Keep reading