Google unveiled a series of artificial intelligence-powered changes to its search engine on Tuesday that fundamentally alter how users interact with what has long been the world’s most-used entry point to the web. The announcement, anchored by a new version of the company’s Gemini AI model, signals a structural shift in a business that has operated largely unchanged for a quarter century.

The redesigned search bar expands dynamically to accommodate longer, more conversational queries, a departure from the compact, keyword-oriented interface that has defined the product since its early years. Users will also be able to upload photographs, videos, and documents directly into search queries, in what Google describes as multimodal functionality.

The most consequential change may not be visual. Google is introducing agentic capabilities directly into search, allowing users to create automated monitors, or “agents“, that track topics, listings, or announcements over time without requiring repeated manual queries. Someone apartment hunting, for example, could configure the tool to alert them when new listings become available, bypassing the need to check real estate platforms directly.

The features are powered by Gemini 3.5 Flash, a new model that Google says performs autonomous tasks faster and at lower cost than comparable systems. According to the New York Times, CEO Sundar Pichai framed the model’s affordability as central to Google’s ability to deploy it at scale, noting that AI-powered search features drive increased overall usage of the platform.

Google is also introducing Gemini Spark, a background agent embedded across Gmail, Docs, and other Google services. Spark can compile notes from across applications, read incoming emails, and generate drafts, capabilities Google described as a direct alternative to autonomous agents offered by Anthropic and OpenAI. Additionally, a revamped shopping experience allows users to build purchase carts while browsing search results or YouTube, with the system flagging incompatible product combinations and surfacing available discounts.

The updates have prompted concern among analysts and researchers about what the shift means for users and for the broader internet economy. Carolina Milanesi, an independent technology analyst, told NPR that the move toward AI-generated results reduces the degree of user choice. “Right now it’s: I ask a question, I get a bunch of answers and I feel that I’m in control as to which answer I take,” she said. “That is going to be less so going forward.”

Sarah T. Roberts, director of the Center for Critical Internet Inquiry at UCLA, raised questions about transparency, telling NPR that algorithmic search results have always been opaque to users, and that adding AI layers will compound that opacity. She pointed to past AI errors (including a notorious instance where Google‘s system advised using glue as a pizza topping) as a caution against moving too quickly.

Financial analyst Richard Kramer of Arete Research, according to the New York Times, offered a more structural critique: “The open web is on its way out. With AI, Google is reducing everyone to raw data providers.” His concern reflects a wider debate about so-called “Google Zero”, a scenario in which AI-generated answers reduce click-through traffic to third-party websites, undermining advertisers, publishers, and online retailers that depend on referred visits.

Google has acknowledged the commercial dimension of the changes. Liz Reid, the company’s head of search, told NPR that longer, more natural queries give Google better insight into user intent (including when a user transitions from research to purchasing) which in turn improves ad targeting. Google’s ad clicks rose six percent last year, with the cost-per-click increasing by seven percent, while the company’s annual profit reached $132 billion, more than double its 2022 figure.

Gemini now counts more than 900 million active users, comparable to ChatGPT’s reported user base, according to multiple accounts of Tuesday’s conference. Despite that scale, Google’s share of paid AI business subscriptions in the United States stood at just 4.5 percent in April, well behind Anthropic at 34.4 percent and OpenAI at 32.3 percent, according to data from financial platform Ramp.