Google Discover has always been one of the quieter power surfaces in Google’s ecosystem. It does not ask you what you want. It watches, infers, and fills the feed.
That may be changing. ZDNET reports that Google is adding a way to tailor Discover using natural language, letting users tell Google what they want to see “in your own words” instead of relying only on the usual algorithmic signals and feedback buttons.
The promise is simple: if your Discover feed is drifting, you can be more direct. Rather than just tapping that you are not interested in a topic, hiding a source, or waiting for the system to learn from your behavior, you can describe the kind of feed you actually want.
Google already offers official ways to customize what appears in Discover, including controls around topics, sources, and activity. The reported natural language layer would make that process feel much less like managing settings and much more like prompting the algorithm.
The feed becomes something you can talk to
This is not just a usability tweak. It changes the relationship between the user and the feed.
For years, algorithmic feeds have been controlled indirectly. You clicked, paused, searched, ignored, followed, unfollowed, liked, or hid. The platform then translated those signals into a prediction about what you wanted next. That made feeds powerful, but also frustrating. When the system got it wrong, users often had to correct it through small, blunt actions.
Natural language control makes that correction more explicit. The user does not have to teach the machine through behavior alone. They can state intent.
That matters because Discover sits somewhere between search and social. It is not a traditional feed of people you follow, and it is not a query box waiting for one answer. It is Google deciding what might be useful before you ask. Giving users a conversational control layer makes that prediction feel more negotiable.
It also makes Discover feel more aligned with the broader AI interface shift happening across Google. As Google’s Gemini app reaches major consumer scale, Google has more reason to train users that instructions, not menus, are becoming the default way to steer digital experiences.
Discovery is becoming a prompt
The bigger signal is that algorithmic control is moving out of the background.
Platforms spent the last decade making feeds feel effortless. The trade-off was opacity. Users got personalization, but not always agency. A natural language Discover control points to a different model: keep the machine learning, but let the user speak more clearly into it.
That is especially important for Google because Discover is not only about content consumption. It shapes attention before intent becomes a search. If users can tell Discover they want more of one subject, less of another, or a different mix entirely, Google gets a cleaner signal than passive browsing alone.
For publishers and brands, the important part is not that Google is adding another control. It is that audience preference may become more actively declared. Content will still need to earn relevance, but the definition of relevance may increasingly come from what people say they want, not just what their clicks imply.
The strategic consequence is clear: the next generation of feeds will not only predict behavior. They will ask users to help steer it.