Build Your Own YouTube Algorithm With AI
▼ Summary
– YouTube announced a new feature called custom feeds that allows users to create personalized video recommendations using natural language prompts.
– The feature utilizes Google’s Gemini AI model to generate and pin specific feeds to the top of the user’s home page based on detailed requests.
– This move aligns with industry trends as other platforms like Bluesky, X, Instagram, and Spotify have also introduced AI-powered feed-building tools.
– YouTube VP Emily Moxley stated that custom feeds aim to help users navigate the platform’s vast library of over 20 billion videos more effectively.
– Support for creating multiple custom feeds will be available on web and mobile starting next month without replacing the main recommendation feed.
YouTube is introducing Custom Feeds, a new feature that empowers users to curate their own video recommendations using natural language prompts. Announced on Wednesday, this tool allows viewers to describe exactly what kind of content they wish to see, leveraging Google’s Gemini AI model to generate a personalized stream of videos. The resulting feed is pinned to the top of the YouTube home page in its own dedicated tab, offering a distinct alternative to the standard recommendation engine.
To create a custom feed, users simply type a description into a prompt box. This can range from specific interests to functional use cases. For example, YouTube suggests building a feed of video podcasts for a 30-minute train commute or requesting “relaxing commentary videos to unwind with.” Because the system is powered by AI, users can provide lengthy, detailed instructions. You can specify what the feed should feature, what to exclude, and which topics to prioritize, giving you granular control over your viewing experience.
This move aligns YouTube with a broader industry trend where social platforms are handing algorithmic control back to users. Decentralized network Bluesky pioneered this concept years ago to distinguish itself from competitors like X. More recently, Bluesky launched Attie, an AI tool that simplifies feed creation. Other major services, including Meta’s Threads, Instagram, X, and Spotify, have also rolled out similar AI-powered features that let users define their feed preferences through descriptive text.
Expanding Discovery in a Vast Library
YouTube aims to help users navigate its massive library more effectively. The platform hosts over 20 billion videos, creating a “treasure trove” of potential content that can be difficult to access without precise navigation. By allowing users to build their own algorithms, YouTube hopes to make discovery more intuitive and tailored to individual needs.
Emily Moxley, VP of Product Management for Viewer AI, emphasized the scale of the platform during a pre-brief ahead of the Made on YouTube event. She stated, “One of the amazing things about being on YouTube is that the world is always at your fingertips. There’s over 20 billion videos in the YouTube corpus, so you’re always one search away from a treasure trove of topics.”
The new custom feeds are designed to complement, not replace, the main recommendation feed. Users will still have access to the traditional algorithm that serves general recommendations. Instead, the custom feeds offer an additional layer of choice, allowing viewers to easily switch between different curated streams depending on their current mood or intent when opening the app.
Availability and Rollout
Support for creating multiple custom feeds will become available across YouTube on web and mobile starting next month. This phased rollout ensures that both desktop and mobile users can experiment with the new AI-driven curation tools. As the feature expands, it represents a significant shift in how users interact with video content, moving from passive consumption of algorithmically determined suggestions to active, self-directed curation.
(Source: TechCrunch)
