Google Changes Discover: You Can Tell AI What You Want to See
Google is testing an AI-powered way to customize Discover with natural-language requests. Here is how it works and what it means for users and publishers.
European Union, 2023 - Source: EP - Margaux D'Herckers
Google is making Discover more conversational
Google is testing a new way to personalize Discover that could change how people interact with the feed. Instead of relying only on signals such as searches, activity, followed interests, and feedback buttons, users can tell Discover in their own words what they want to see.
The experiment is called “Tailor your feed.” It lets users describe topics, sources, formats, moods, goals, and even the kinds of recommendations they no longer want. Google’s AI can then update the Discover feed and remember those preferences so they can be adjusted later.
That is an important shift. Discover has traditionally tried to predict what may interest a user before the user performs a search. With this experiment, personalization becomes more explicit and conversational.
What is “Tailor your feed”?
According to Google’s Search Labs description, the feature allows users to tell Discover what they want to see in natural language. The requests can be very specific or broad.
For example, a user could ask Discover to:
- Add a new subject, such as training advice for a first half marathon.
- Keep up with a particular area of news or sport.
- Stop showing travel recommendations after returning from a trip.
- Reduce a category of content they no longer want.
- Refine an existing interest, such as seeing dairy-free recipes.
- Show more content from a preferred publisher.
- Prefer a format such as meal-preparation videos.
- Change the overall feel of recommendations to something calmer or more focused.
Google says the AI can update the feed immediately and remember the user’s preferences. The controls are not necessarily permanent: preferences can be changed again whenever needed.
How the new AI personalization works
The experience is designed around a simple idea: describe the outcome you want instead of trying to train the feed one card at a time.
A user may not know which topic labels or settings would produce the right result. Natural-language instructions make that process easier. Someone can explain a project, a temporary interest, or a change in circumstances, and Discover can use that instruction as an additional personalization signal.
The feature can be accessed through Discover when available, including from the more-options menu on a recommendation. Because it is a Search Labs experiment, availability is limited and may change.
At the time of writing, Google’s official experiment page says it is available only to users aged 18 or older in the United States, in English, through the latest Google app on a phone. Space is also limited, so the feature should not be treated as a global rollout.
What changes for everyday Discover users?
The biggest potential benefit is control.
Traditional recommendation systems can misunderstand short-term behavior. Reading one article about a destination, a sport, or a product can sometimes influence later recommendations more than a user expects. A conversational control gives people a clearer way to correct the feed.
It may also make Discover more useful during projects and life events. A person planning a trip, learning a skill, preparing for an exam, or following a temporary project can explain that context directly.
Google already provides other ways to influence Discover, including following or unfollowing interests and managing personalized recommendations. The AI experiment does not replace the need for privacy and personalization controls. Users should still review the Google account settings that affect saved activity and personalized recommendations.
What does this mean for publishers and SEO?
For publishers, the change reinforces a trend that has been visible across Search and Discover: relevance is becoming increasingly personal.
A good headline alone may not be enough to explain why a piece of content matters to a particular reader. If users can explicitly ask for more coverage of a subject, source, format, or style, publishers have an additional reason to build recognizable topical coverage and clear editorial identities.
This does not mean publishers can prompt their way into a user’s feed. Google has not announced a new ranking shortcut or a guaranteed placement system for “Tailor your feed.” The safer interpretation is that strong, relevant content may have more opportunities when it matches preferences a user has deliberately expressed.
For SEO teams, several principles remain important:
- Cover topics with genuine depth rather than publishing isolated articles.
- Use clear headlines and descriptions that accurately communicate the subject.
- Build consistent expertise in identifiable topic areas.
- Maintain trust, original reporting, and useful updates.
- Make visual and mobile experiences strong enough for Discover consumption.
- Develop a recognizable publisher identity that readers may actively choose to follow or request.
Featured Snippet Answer: How can you customize Google Discover with AI?
Google’s “Tailor your feed” experiment lets eligible users describe the content they want to see in natural language. You can ask Discover to add or remove topics, refine recommendations, prefer certain publishers or formats, or match a particular mood. Google’s AI then updates the feed and can remember those preferences.
Frequently Asked Questions
Is Google Discover getting an AI chatbot?
Google is testing a conversational AI interface for feed personalization. The current official experiment is called “Tailor your feed,” and it is designed to change recommendations based on natural-language requests.
Can I tell Discover to stop showing a topic?
Yes. Google’s examples include asking the feed to remove content that is no longer useful or wanted.
Is the feature available everywhere?
No. Google’s official experiment page currently lists limited availability: users 18+, English, the United States, and the latest Google app on a phone.
Will publishers be able to force their articles into Discover?
There is no official indication that this feature creates a new guaranteed placement mechanism. Content still needs to be relevant to the user and Discover’s broader systems.
Does AI personalization replace Discover’s existing controls?
No. Google still provides settings and controls for interests, saved activity, and personalized recommendations.
The bottom line
Google’s experiment turns Discover personalization from a mostly implicit process into a more direct conversation. Users may no longer need to wait for the system to infer every interest from clicks and searches; they can explain what they want.
For publishers, the message is equally clear: the future of discovery may depend even more on being useful to a specific audience with a specific interest. As Discover becomes more conversational, clear topical expertise, recognizable editorial value, and content that genuinely satisfies reader intent could become even more important.
Related Articles
- Agentic AI Explained: How Intelligent Agents Are Transforming Work
- OpenAI Hacks 2026: Pro Secrets to Improve ChatGPT
