YouTube's custom feeds let users design their own recommendation algorithm
YouTube is introducing custom feeds that allow users to describe the types of videos they want in plain language, with Gemini AI generating a tailored feed. The resulting feed appears as a pinned tab on the home page. This feature follows similar AI-powered feed-building tools from other platforms like Bluesky.
The feature was unveiled at the Made On YouTube event on Wednesday, where the company demonstrated how users can type a detailed prompt—such as a feed for a specific commute time or mood—and Gemini AI will curate a dedicated tab pinned to the top of the home page. This custom feed operates separately from the main recommendation stream, allowing users to jump between different curated views.
YouTube's VP of Product Management for Viewer AI, Emily Moxley, highlighted that the platform hosts over 20 billion videos, making discovery a key challenge. The rollout begins next month across web and mobile, with support for multiple feeds. This follows similar AI-powered feed-building tools from Bluesky's Attie, Threads, Instagram, X, and Spotify, signaling a broader industry shift toward user-defined algorithmic control.
This feature could give viewers unprecedented control over their content discovery, potentially reducing reliance on opaque, engagement-driven algorithms. However, it may also encourage users to retreat into narrow interest bubbles, limiting serendipitous discovery. For creators, visibility could become more fragmented as audiences self-segment into custom feeds, possibly affecting ad revenue and growth strategies. The broader trend toward user-customizable feeds across platforms suggests a future where algorithmic curation becomes more transparent and intentional, though its long-term effects on platform engagement and content diversity remain uncertain.