Google adds AI chatbot to personalize Discover feed

Google announced that its Discover feed will soon include a new AI‑powered customization option. Users will be able to open a chatbot‑style prompt from the three‑dot menu on the feed and type in the kinds of stories or topics they wish to prioritize. After the user’s input, the chatbot confirms the selection, outlines the content categories it will favor, and offers a chance to refine the request if the initial interpretation is off. A final "Refresh your feed" button applies the changes, and the system stores the preferences for subsequent sessions.

The feature is slated to appear in the Google app within the next few days, according to a video shared by the company. It operates by interpreting natural‑language descriptions and adjusting the recommendation algorithm accordingly, effectively turning a traditionally opaque ranking system into a more transparent, user‑controlled experience. The interface also allows users to add extra details if the AI does not capture their intent perfectly, aiming to reduce the friction of manual feed tweaking.

Discover is Google’s personalized news surface that surfaces articles, videos and other content based on a user’s search history, app usage and broader activity across Google services. Historically, the feed has been driven by algorithmic signals that users could not directly influence, leading to occasional complaints about irrelevant or repetitive suggestions. By giving users a conversational way to steer the feed, Google hopes to make the experience feel more intentional and less serendipitous, aligning the content stream with explicit user interests.

Several social platforms have already introduced AI‑assisted feed controls. YouTube lets creators and viewers fine‑tune recommendations, Instagram offers AI‑generated explore suggestions, while newer services like Bluesky and X have experimented with prompt‑based personalization. Google’s move places its flagship feed alongside these competitors, but it distinguishes itself by embedding the customization directly within the native app and leveraging its extensive data ecosystem to interpret user prompts.

The rollout is not without potential drawbacks. The AI must accurately parse natural language, and early iterations may misinterpret vague requests, requiring users to iterate through the refinement step. There are also privacy considerations, as the system will need to retain preference data tied to a user’s profile. Moreover, the feature’s impact on content diversity remains unclear; overly narrow preferences could limit exposure to a broader range of information.

Alongside the Discover update, Google revealed additional personalization tools. Android users of the Google News app can now tailor daily audio briefings to focus on chosen topics, and the Preferred Sources feature has been upgraded to let publishers embed an interactive button that adds their outlet to a reader’s favorites without leaving the page. These changes collectively signal Google’s broader strategy to give users more granular control over the information they consume across its ecosystem.

The chatbot customization will begin rolling out in the coming days, initially to a subset of users before expanding globally. Observers will watch adoption rates, the accuracy of the AI’s interpretations, and any shifts in user engagement metrics. Future updates may refine the natural‑language model, integrate more granular category selections, or extend the approach to other Google surfaces such as Search or YouTube recommendations.