Meta claims that this update will make Facebook more personal and vibrant, where users will be more connected to friends and can easily find the content they want.
Meta announced on Tuesday that its new Reels algorithm will now prioritize new and recent videos, meaning users will see 50 percent more new Reels as they scroll through their feeds than before. It also added an artificial intelligence (AI)-powered recommendation tool that will curate content based on user interests.
"If a user shows interest in AI-generated content, the system will understand that and show more of the same type of content. If someone doesn't like it, the algorithm will quickly recognize that and show less of that type of video," said Jagjit Chawla, vice president of product at Meta.
The new interface includes a simple 'Not Interested' button, video saving, and personal collection organizing. Users can now curate their own viewing experience, instead of relying solely on algorithms.
The Reels feed will now focus on “what’s happening right now.” That is, new, recent, and timely content will be more visible. Similar to TikTok, AI-powered search suggestions will now appear below the video. Where keywords will be suggested based on the user’s watched videos and interests. This will make content search easier.
Chawla said, "We want Facebook to become a network for people again. Where friends' activities are visible."
Facebook's new initiative shows how deeply generative AI is now embedded in social media. Chawla confirmed that videos created with tools like OpenAI's Sora or Midjourney will be treated the same as human-made videos by Facebook's algorithm.
Facebook says its AI system now also gives special weight to users' negative feedback clicks, so that unwanted content can be quickly removed.
With this update, Meta is making it clear that they are not only working to increase video watch time but also to restore social connections between people.
According to Facebook, this new Reels experience will make the platform more lively, relevant, and user-centric.
