Facebook is implementing a significant upgrade to its Reels algorithm, aiming to enhance user experiences by making them more dynamic, timely, and personalized. This update, announced by Jagjit Chawla, Facebook’s Vice President of Product, signifies another step in Meta’s initiative to leverage artificial intelligence to enhance content visibility on users’ feeds. With video watch time on Facebook increasing by 20 percent year over year, Meta is reevaluating its methods for connecting users with relevant and up-to-date content. The revamped Reels will now prioritize displaying 50% more videos that were posted on the same day users browse, thereby placing a greater emphasis on timeliness and trending topics.
Chawla noted that this change is in response to user feedback calling for more recent and significant content. “Our internal testing indicates that this strategy boosts engagement,” he stated, adding that the aim is to make Reels feel more immediate and relatable. In addition to the algorithmic update, Facebook is launching AI-driven search suggestions within Reels, akin to a feature seen at the bottom of TikTok videos. These suggestions will assist users in discovering new creators and topics based on their past interactions and viewing patterns. Another feature, termed “friend bubbles,” is designed to enhance the social aspect of the experience.
It allows users to send quick direct messages to friends who have liked a specific Reel, facilitating conversations and interactions around shared content. Currently, there is no dedicated feed for friends’ Reels, so this feature addresses that gap by promoting spontaneous engagement within the app. However, the increasing prevalence of AI-generated content has sparked new concerns regarding authenticity, misinformation, and copyright issues. As tools like OpenAI’s and Midjourney’s generative models make it easier to produce realistic videos, Meta is striving to assist users in differentiating between human and AI-generated content. The company provides both manual and automatic labeling options for AI-generated posts, although it acknowledges that detection systems are not foolproof.
Chawla emphasized that Facebook’s recommendation engine treats both AI-generated and human-created videos equally. “If you, as a user, show interest in a piece of content that happens to be AI-generated, the recommendations algorithm will learn over time that you are interested in that topic and content,” he explained. Conversely, when users indicate they are “Not Interested” in something, that feedback significantly impacts what appears in their future feeds. He also pointed out the evolving nature of content formats and how AI streamlines the process for creators and users alike. “Technology will blur the lines between whether it was long or short [form video], horizontal or live, or any of that,” Chawla remarked.
“When I engage with both users and creators, I [say], pursue your creativity, and the responsibility for execution lies with technology, which we will manage.” As Meta continues to rely on AI, the company’s objective remains clear—empowering creators to produce authentic, high-quality videos while ensuring that users receive the most engaging and relevant content on their feeds. Video, Chawla affirmed, will remain central to Facebook’s growth strategy.