Google changes search box for AI

Google used its May 19 developer conference to change the shape of its core search product. At I/O 2026, the company said Search would take longer, more conversational prompts as it pushed users toward AI Mode and other Gemini-powered features. Google also introduced Gemini Spark, a more proactive assistant inside the Gemini app, and Gemini Omni, a model built for generating and editing content including video. (blog.google 1) (blog.google 2) The changes put Google’s best-known interface — the search box — at the center of its AI push. They also arrived as Anthropic added Andrej Karpathy, a former Tesla AI executive and OpenAI founding member, underscoring how product competition and recruiting are moving in parallel. (blog.google) The New York Times reported that Google is redesigning the search box for the first time in 25 years and that Gemini has about 900 million regular users. ### What did Google actually change in Search? Google said on May 19 that it was entering “a new era for AI Search” and expanding a search experience built around longer, more complex questions. In earlier updates this year, Google said users could ask “whatever’s on your mind — no matter how long or complex” and move from AI Overviews into a back-and-forth conversation in AI Mode. (blog.google) Search has long been built around short keyword queries. The New York Times reported that Google is now changing the dimensions and function of the search box itself to better fit AI-style prompts, while also adding shopping and video features tied to Gemini. (blog.google) ### Where does Gemini Spark fit into this? Google said Gemini Spark is part of a Gemini app that is becoming “more agentic” and more proactive. In its I/O roundup, the company said Spark and Daily Brief are examples of agentic experiences it is adding across products. Sundar Pichai said in his keynote transcript that Google is in a phase where people want to see AI in products they use every day. (blog.google) Google framed the new tools as moving beyond systems that help users write toward agents that help users act. ### What else did Google launch alongside the search-box redesign? Google said Gemini Omni can “create anything from any input, starting with video,” making video generation one of the headline product additions at I/O. (blog.google) The company also introduced Gemini 3.5 Flash, which it described as the first release in a model family built for more capable agents and coding tasks. Google also used the event to promote shopping tools. Its I/O collection page highlighted Universal Cart, which it described as “a truly intelligent shopping cart,” alongside Search and Gemini updates. (blog.google) ### How broad is Google’s AI distribution now? Google said Gemini 3.5 Flash is available “today to billions of people globally” through the Gemini app and AI Mode in Google Search, as well as through developer and enterprise products. That reach reflects Google’s advantage in distribution as it folds AI into products that already have large audiences. The New York Times reported that Gemini now has about 900 million regular users. (blog.google) Google did not provide that figure in the official materials reviewed here, but its product announcements repeatedly emphasized deployment across Search, the Gemini app and other consumer services. ### Why does Anthropic’s Karpathy hire belong in this story? Anthropic announced on May 19 that Andrej Karpathy had joined the company, according to Reuters. Karpathy is a former Tesla AI executive and an OpenAI founding member, making the hire notable in a market where a small number of labs are competing both for users and for senior technical talent. (blog.google) Google’s I/O announcements and Anthropic’s hiring move happened within a day of each other. Together, they show the AI contest playing out through consumer product rollouts on one side and researcher recruitment on the other. That framing is an inference based on the timing and substance of the two developments. (blog.google) ### What comes next from here? Google said I/O 2026 runs on May 19 and May 20, with additional developer sessions and product details released through its conference collection page. The company said Gemini 3.5 Pro is expected next month, while AI Search, Gemini Spark and shopping features are being rolled out through existing Google products. (developers.googleblog.com) (blog.google)