Google is testing a new “Dive deeper” feature in Discover that uses AI to give users more context around a topic before they move on to other content.
The experiment was announced by Google Search VP Robby Stein and is starting with videos. When a user taps Dive deeper, Google opens an AI-generated topic overview containing links to related stories, community reactions and original reporting. Different designs will be tested over the coming weeks.
It sounds like a fairly small Discover experiment. For publishers, though, it changes something more significant: what can happen between discovering a story and clicking through to its source.
How ‘Dive Deeper’ Works
Google’s example starts with a weather video about a Northern California heat wave. Beneath it is an AI-generated summary carrying a warning that the information can contain mistakes. Tap Dive deeper and the experience expands into a longer overview, source cards and an “Explore more stories” section containing additional coverage of the topic.
So this isn’t simply a summary of the video. Google is building a broader information layer around the subject, giving someone a way to continue exploring it without returning to Discover or running a separate search.
The test begins with video content. Google hasn’t specified the countries, languages or number of users involved, so this remains an experiment rather than a Discover-wide rollout.
Discover Is Moving Further Into AI
This isn’t Google’s first attempt to connect Discover with generative AI. It has previously experimented with sending Discover users into AI Mode, where they can continue exploring a subject through generated responses.
When Google introduced Discover, the basic idea was to surface useful content based on a person’s interests without requiring a search query. Dive deeper expands that role. Instead of only recommending content, Discover can now generate context around a subject, connect related coverage and help users explore it further. It fits with Google’s broader push to make generative AI a more prominent part of how people explore information in Google Search.
Discover is starting to do more than recommend what to read next.
What It Could Mean for Publishers
The obvious question is traffic. Discover can provide publishers with a relatively direct route from a recommendation to their website; Dive deeper introduces an AI-generated destination before a user necessarily visits another source.
That isn’t automatically bad for publishers. A site whose reporting wasn’t featured in the original Discover card could gain exposure if Google selects it as a source inside the expanded overview. Someone might begin with one publisher’s video, for example, then encounter reporting from another publication after tapping Dive deeper.
But the opposite possibility also, is difficult to ignore. If the AI overview provides enough context to satisfy someone’s curiosity, there may be less reason to open one of those external sources. Google already provides publishers with a dedicated Search Console report for generative AI features in Discover, but an experiment like this makes understanding what happens after an impression increasingly important.
We don’t yet know which behaviour will dominate.
Where Dive Deeper Could Take Discover
The interesting part of Dive deeper isn’t another AI summary. Google already uses generative AI extensively across Search.
It is where the summary appears that makes the experiment worth watching. Someone can discover a video, move into an AI-generated overview, browse related reporting and then decide whether to visit a publisher. There is no conventional search query anywhere in that sequence. For publishers, the question is fairly simple, even if the answer isn’t: does being included inside these AI-generated Discover overviews create another opportunity for a click, or another place where the click disappears?
