Facebook is trying to help you see the Reels videos you’re actually interested in, rather than random videos. A new algorithm update will show newer Facebook Reels. Along with a new AI-powered search tool, Facebook said it wants to return to its roots to connect you with friends and family.
The algorithm update will prioritize newer content, showing you 50% more Reels that were posted on the same day you’re scrolling. In an exclusive interview about the new recommendation engine changes, Facebook’s vice president of product, Jagjit Chawla, told CNET that Facebook users have been requesting their feeds show them more recent videos, and Facebook’s internal testing found that doing so would help keep users returning to the app. Video is one of the most common ways people use Facebook, with the company reporting that time spent viewing video has grown 20% year over year.
Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.
Facebook Reels is also adding AI-powered search suggestions. These are similar to the gray search suggestions that appear at the bottom of TikTok videos. This is different from Meta’s news earlier this month that it will use your AI interactions for personalized ads and content recommendations beginning on Dec. 16. These search suggestions are powered by AI and machine learning tech behind the scenes that recommends search queries based on interests you’ve shown.
Facebook Reels are also getting friend bubbles similar to the ones on Instagram — those little icons of your friend’s profile picture that float in the lower left corner of posts they’ve liked. You can tap their profile picture and instantly start a DM. This feature could be especially helpful for Facebook Reels users, as there isn’t a dedicated Reels feed of your friends’ videos, unlike on Instagram.
AI in our feeds vs. behind the scenes
Chawla said AI-generated content, like videos created by OpenAI’s Sora or Midjourney, is treated the same as human-generated content when it comes to serving them in users’ Facebook Reels feeds.Â
“If you, as a user, are interested in a piece of content which happens to be AI-generated, the recommendations algorithm will determine that, over time, you are interested in this topic and content,” Chawla SAID. “If you are not into it, which, for lack of a better term, there is a set of users who would consider that content AI slop, and if you have given us signals that this is not for you, that algorithm will respond appropriately to make sure we don’t show you more of that.”
AI-edited or created content has become more commonplace on social media platforms, thanks to the rise of AI image and video generators. Meta, which owns Facebook and Instagram, has been quick to integrate AI on its social platforms, to the dismay of many users. Generative AI companies like Midjourney and OpenAI have also made it easier and faster than ever to create realistic-looking content, sparking concerns about misinformation and copyright infringement. On Meta’s platforms, users have the option to add labels to AI-generated content. Meta also has its own labels. But AI detection is far from perfect.
If you don’t want to see AI content, it’s important to send those negative signals to platforms. That’s true for any content you don’t want to see, too. Chawla said from an algorithm point of view, users tend to give more positive signals, by liking, commenting or sharing. So when Facebook does receive a negative signal — you tapping the “Not Interested” button, for example — Facebook “take[s] that a lot more seriously from an algorithm standpoint,” Chawla said.
For a long time, creators trying to hack the algorithm have had to quickly adapt to an ever-changing set of rules to ensure their content makes it into our feeds. But with more AI-powered automation working behind the scenes to create our feeds and growing masses of AI content, ultimately, the content that’s most likely to break through the noise is authentic and human-driven.Â
“Technology will obviate the lines between whether it was long or short [form video] or horizontal or live, or any of that,” Chawla said. “When I talk to both sides, users and creators, I [say], do your thing, and the responsibility and the heavy lift is on technology that we will figure out. As AI advances, it’s just making our job easier.”