YouTube to Let Users Build Their Own AI-Powered Algorithm

YouTube is giving viewers a new way to shape what they watch. The platform announced a feature called custom feeds that lets users describe, in plain language, the type of videos they want recommended — a step toward letting people build their own recommendation algorithm rather than relying solely on YouTube's default one.

To create a custom feed, users type a description into a prompt box on YouTube. The feed can focus on a specific interest or serve a particular purpose. YouTube's own examples include a feed of video podcasts sized for a 30-minute train commute, or mood-based requests such as "relaxing commentary videos to unwind with."

Powered by Gemini

The prompt-based feature runs on Google's Gemini AI model, which builds the feed around each user's request. Once generated, the feed is pinned to the top of the YouTube home page in its own tab. Because the creation process is AI-driven, prompts can be as detailed as users like — covering how the feed should be curated, what content to feature or exclude, and what to prioritize.

Once live, custom feeds will not replace YouTube's main recommendation feed. Instead, users will be able to jump directly into whichever feed fits their current needs when they open the app.

Part of a Broader Trend

The move follows similar efforts across the social media industry, where platforms are increasingly letting users describe what they want to see rather than relying on one black-box algorithm.

Decentralized network Bluesky helped popularize the concept years ago as a way to set itself apart from rivals like X, and this year it introduced an AI tool, Attie, that makes feed-building even simpler. Meta's Threads, Instagram, X, and more recently Spotify have all launched their own feed-customization features, many of them AI-powered as well.

Why It Matters

YouTube frames the feature as a way to help users navigate the enormous scale of its catalog. "One of the amazing things about being on YouTube is that the world is always at your fingertips," said Emily Moxley, VP of Product Management for Viewer AI, during a pre-brief with reporters ahead of Wednesday's Made on YouTube event, where custom feeds were introduced to the public. "There's over 20 billion videos in the YouTube corpus, so you're always one search away from a treasure trove of topics."

Support for creating multiple custom feeds will roll out across YouTube on web and mobile starting next month.

Meta description: YouTube's new custom feeds use Gemini AI to let users describe the videos they want recommended, pinning personalized feeds to their home page.

Tags: YouTube, Gemini AI, custom feeds, recommendation algorithm, streaming

Featured image: Abstract technology style — a stylized video grid dissolving into flowing streams of light, no people, logos, or text.