YouTube lets US users build personalized video feeds with conversational AI

Yesterday 23:55

YouTube is giving viewers a more direct way to shape what appears on their home page, using a conversational artificial intelligence tool that lets users describe the content they want to discover.

The feature, currently available to signed-in users in the United States, introduces a new “Your custom feed” option on YouTube’s home page. Rather than relying entirely on the platform’s recommendation system, viewers can use natural language to tell the AI what subjects, formats or types of videos they would like to see.

A conversational approach to recommendations

The new interface works much like a conventional AI assistant. Users can begin with a simple request and then refine it through additional prompts, allowing the system to adjust the selection of recommended videos.

For viewers with an existing YouTube history, the tool can also draw on previous activity to suggest possible interests. This creates a more interactive recommendation process, combining information already known about a user with preferences expressed directly in conversation.

Once a personalized feed has been created, users can save it and return to it later without having to repeat the original request. YouTube also allows people to create more than one custom feed, potentially making it possible to maintain separate recommendations for different interests.

Moving beyond passive personalization

YouTube has long relied on viewing history, subjects watched and preferred formats to determine which videos appear in users’ recommendations. The new system adds another layer by allowing viewers to communicate their preferences explicitly.

The change reflects a broader shift in how digital platforms are using generative AI. Instead of simply presenting algorithmically selected content, platforms are increasingly experimenting with interfaces through which users can describe what they want and receive dynamically adjusted results.

For YouTube, the approach also strengthens the role of AI within a service that already processes enormous amounts of video and viewing data.

AI expands across YouTube

The personalized-feed feature is part of a wider effort by YouTube to incorporate artificial intelligence into both content consumption and production.

In 2025, the platform introduced several AI-powered creative tools, including video generation, photo animation and the ability to insert objects or characters into videos. These features have expanded the range of ways creators can experiment with automated production tools.

At its annual Made on YouTube event, the company also presented further developments aimed at changing how audiences interact with video.

Among them is real-time dubbing for live broadcasts, scheduled to arrive in early 2027. The technology is intended to make live content more accessible to audiences speaking different languages.

“Ask YouTube” adds another AI layer

YouTube is also introducing “Ask YouTube,” a generative AI interface designed to answer users’ questions about subjects they are interested in.

Instead of returning only a conventional search-results page, the tool generates a short written response and recommends several videos connected to the question. This places conversational assistance directly alongside YouTube’s existing video-discovery system.

Together, the new tools point toward a more interactive model for navigating the platform, where users can increasingly describe their interests in ordinary language rather than relying exclusively on searches, subscriptions or automated recommendations.

For now, the custom-feed function remains limited to signed-in users in the United States, leaving its broader international availability to a future stage.