For nearly three decades, Google’s search box did one thing extremely well. You typed a few words, it handed you ten blue links, and you clicked through to find your answer somewhere else. That basic contract, type a query, get a list, go elsewhere, is the thing Google just tore up. At I/O in May, the company rolled out its biggest search overhaul in 25 years, and the Google AI search changes at the center of it go well beyond a new coat of paint.
The headline shift is technical but consequential: AI Mode is now running on Gemini 3.5 Flash, Google’s newest model, and it’s the default experience for AI Mode globally, not an opt-in feature buried in a menu. According to Google’s own announcement, AI Mode surpassed one billion monthly users within a year of launch, with query volume more than doubling every quarter since. That’s not a niche feature anymore. That’s becoming the main way a meaningful share of the internet searches.
The search box itself changed too, and this is where the Google AI search changes become impossible to ignore even for casual users. The input field that has barely moved since 1998 now accepts text, images, files, video, and even open Chrome tabs. It expands dynamically as you type, and it offers AI-powered suggestions that go well past autocomplete, actively helping you formulate the question you’re struggling to articulate. Liz Reid, who leads Search at Google, called it the biggest upgrade to the search box since its debut. That’s not marketing hyperbole so much as a fair description of what changed under the hood.
Underneath the new box sits a fundamentally different interaction model. Instead of a static list of results, a query now triggers a synthesized answer generated by Gemini 3.5 Flash, which then opens into a conversation. You can ask follow-up questions directly beneath the answer, refine what you meant, or pivot entirely, all without leaving the page or starting a new search. Reporters covering the announcement compared the new layout to opening a chatbot rather than a search engine, and that comparison holds up. The lines between “searching” and “chatting with an AI” are blurring fast, and Google, which controls the front door to most of the web, just accelerated that blur deliberately.
Then there are Information Agents, arguably the most consequential piece of the announcement even though it got less attention than the redesigned box. These are persistent, background processes that monitor the web on a user’s behalf, continuously, without being asked twice. Tell it you’re apartment hunting with a specific set of requirements, and it will keep scanning listings and notify you when something matches. Tell it you want to know the moment a favorite athlete announces a new shoe collaboration, and it will watch for that too. Google says these agents pull from blogs, news sites, social posts, and real-time data feeds covering finance, shopping, and sports, then send a synthesized update with the option to act on it directly. Information Agents are rolling out first to Google AI Pro and Ultra subscribers this summer, with a broader release expected later.
Google also introduced Universal Cart, a cross-platform shopping tool that lets people buy things through Search, Gemini, YouTube, and Gmail without ever visiting a retailer’s website, along with expanded agentic booking for restaurants and local services. Put together, these features point toward a search experience that doesn’t just find information for you, it increasingly acts on it, completing tasks that used to require several separate visits to several separate sites.
All of this raises an obvious question for anyone who runs a website, publishes content, or depends on search traffic to reach an audience: if Google is answering questions directly inside the results page and increasingly completing transactions without sending users anywhere else, what happens to the sites that search used to send people to?
The early data is not comforting for a lot of publishers. Independent SEO analysis following Google’s May 2026 core update found that roughly 80 percent of top three search results shifted position, and around 24 percent of pages that had been ranking in the top ten fell out of the top hundred entirely, a sharper disruption than the update that preceded it. The directional trend is clear even if the exact numbers vary by analyst: visibility is moving away from aggregator and summary sites and toward brands, official sources, and destinations with genuinely original, data rich content. Google has been explicit that it still considers this SEO, just applied to a different surface, and the company has pushed back on the idea that publishers need an entirely separate strategy for AI visibility versus traditional search visibility. Whether that reassurance holds up as AI Mode keeps growing its share of total queries is the open question every publisher is now watching closely.
There’s a reasonable counterargument to the doom narrative that’s worth taking seriously. Google has repeatedly framed these Google AI search changes not as an abandonment of the open web but as an evolution of the same underlying goal, helping people find genuinely useful information, just through a richer interface. The company has also rolled out new Search Console reporting that shows impressions specifically within AI Overviews and AI Mode, giving publishers actual visibility into how they’re performing on the new surface rather than leaving them guessing. And Preferred Sources, a feature that lets users explicitly choose which publishers they want prioritized, has been expanded into AI Overviews too, which gives loyal, engaged audiences a direct lever to keep favorite sources visible even inside AI generated answers. None of that undoes the disruption, but it does suggest Google has some interest in keeping the broader publisher ecosystem viable, if only because Gemini’s answers are only as good as the web content it’s drawing from.
Still, the incentive structure is shifting in a way that’s hard to spin as neutral. When Google’s CEO discussed the changes in a podcast interview recorded shortly after the announcement, he acknowledged that AI Overviews can be “more opinionated” than a simple list of links, surfacing a single recommendation where a traditional results page might have shown competing perspectives from multiple sources side by side. That’s a meaningful change in how information gets filtered before it reaches a user, and it concentrates a lot of interpretive judgment inside one company’s model rather than leaving that judgment to the reader comparing sources themselves.
What seems most durable about this moment is the underlying behavioral shift, not just the specific features. Search queries are getting longer and more conversational. People are attaching images, documents, and video to their questions instead of typing three keywords and guessing. They’re asking follow-ups instead of starting over. That’s a genuinely different relationship with a search engine than the one most of us grew up with, and it’s not something Google can easily walk back even if a future update softens some of the rougher edges around publisher traffic.
Whether these Google AI search changes ultimately make the web more useful or simply more centralized around one company’s synthesis of it is going to depend on decisions Google hasn’t fully made yet, around how much traffic still flows outward, how transparent the new reporting tools actually stay, and how much room remains for readers to compare sources rather than accept a single AI generated answer. The technology shift already happened. What it means for everyone downstream of it is still being written.
