YouTube will let you build your own algorithm with AI

You can now tailor YouTube's video recommendations to your specific moods and activities with custom AI-generated feeds.

YouTube is introducing a new feature called custom feeds, which will allow users to build their own recommendation algorithms using natural language prompts. Leveraging Google's Gemini AI model, users can describe the type of videos they want to see, such as a feed for "video podcasts to watch during my 30-minute train commute" or "relaxing commentary videos to unwind with." These custom feeds will be generated based on detailed descriptions provided by the user, specifying content to include or exclude and prioritization. The new feeds will appear as a separate tab at the top of the YouTube homepage and will not replace the main recommendation feed. This feature, rolling out next month on web and mobile, follows similar user-driven feed customization tools launched by platforms like Bluesky, X, Threads, Instagram, and Spotify. YouTube aims to help users better navigate its extensive library of over 20 billion videos.

This move signals YouTube's strategic embrace of generative AI to deepen user engagement and provide more personalized content discovery, directly competing with other platforms that have empowered users to curate their own algorithmic experiences. By allowing granular control over recommendations through natural language prompts, YouTube is democratizing algorithmic customization, making sophisticated content filtering accessible to a wider audience and potentially setting a new standard for user-driven feed creation on large-scale video platforms. The integration of Gemini AI aims to make this process intuitive, even for complex requests, thereby enhancing user satisfaction and retention by ensuring a more relevant viewing experience.

The introduction of custom feeds on YouTube indicates a broader industry trend toward user agency in algorithmic content delivery, moving beyond passive consumption to active curation. Competitors like Bluesky, X, and Spotify have already introduced similar features, highlighting a growing demand for personalized and controllable content streams. YouTube's adoption of this trend, powered by its vast video library, could significantly influence how users interact with video content, making it easier to find niche interests or specific types of videos on demand. This feature could also serve as a valuable tool for creators, offering insights into what specific user-defined feeds are prioritizing.