Meta has driven a double-digit increase in the time users spend on Instagram, thanks to next-generation AI recommendation algorithms deployed across its feed and Reels content. According to the company's financial results, systems that analyze the tone and subject matter of posts using large language models are directly increasing user engagement and session durations.
How AI Analyzes Posts
Meta's updated infrastructure does not evaluate content solely based on watch time or like counts. According to Chief Financial Officer Susan Li, the company has trained large language models to directly grasp the meaning, subject, and tone of every single Reel and feed post.
This semantic comprehension process is paired with the user's past viewing habits to predict with high accuracy what they will want to see next. In addition to automatically processing all public posts on Instagram through this LLM pipeline, a specialized family of models called "Muse" is actively used for topic classification and summarization tasks.
Industry Implications and Effects on the Platform
This major Reels ranking algorithm update delivered a 15-basis-point increase in Instagram sessions, with the greatest gains seen in reshares and total platform usage time. Carrying this success from Reels over to the main feed has allowed users to stay in the feed longer, while also creating more engagement opportunities for advertisers.
Frequently Asked Questions
What exactly does the "Muse" model family used by Meta do within the system?
Muse is a distinct family of AI models deployed to manage the automated topic classification and summarization of content on the platform.
What kind of opportunities does this AI integration create for advertisers?
The increase in the total time users spend on the app and session frequency generates more ad inventory and reach opportunities on the platform.
*This news is based on data published by Webtekno — Artificial Intelligence.
💬 Comments
No comments yet. Be the first!
You must be logged in to comment.
🔑 Log In