The artificial intelligence agents of the Google web will bring you now

Over the past two decades, Google has brought people a list of links chosen from the web an algorithm for any specific research inquiries. In I/O 2025, Google explained that the concept of research is firmly in the rear vision mirror.

On Tuesday, the CEO of Google Sundar Pichai and its CEOs presented new ways to bring users to the web, this time it mediated through a series of artificial intelligence agents.

“We cannot be more enthusiastic about this chapter than Google’s search where you can really ask anything (…) the simplest and most difficult questions, the deepest your search, your custom shopping needs,” said VP of SEARCH from Google, Liz Reid, Onstage in I/O. “We believe that artificial intelligence will be the most powerful engine to discover the web watch ever.”

That was the biggest advertisement of i/o Google now offers artificial intelligence mode for every search user In the United States. This gives hundreds of millions of people a button to talk to the artificial intelligence agent visits web pages, or summarizes them in any way they want, or even Help them shopping. With Project Mariner, Google offers Even more artificial intelligence agents for its subscribers is superior. This agent will deal with 10 different tasks at one time, visit web pages and click on those pages while users are free to connect something else completely.

Google also does it Deep search agentThat visit dozens of relevant websites and create comprehensive research reports, more customized and connect to Gmail and Drive. In the development of parallel, the company More integration of the ASTRA project -The multimedia experience experience in the actual time-in research and criminals, which gives users more ways to speak orally with the artificial intelligence agent and allow it to see what they see.

I can continue, but you get the idea – artificial intelligence agents control I/O 2025.

He was forced to climb Chatgpt at the expense of artificial intelligence in Google, causing rethinking how to bring users from the web from the web. This account has already started in I/O for the past year, when Google I presented an overview of artificial intelligence In the search, his release by it Embarrassed hallucinations. Amnesty International’s overview of Amnesty International looks like artificial intelligence was not ready for primitive time, and this research as we know was here to survive.

But in I/O 2025, Google made a more persuasive approach, embodied in how artificial intelligence reshapes, and thus the web. The new company’s vision indicates that the future of the web and the company includes artificial intelligence agents who bring information from the web and provide it to users in any way they want.

The idea that the agents of artificial intelligence in Google can solve the subject of research is a convincing idea, especially since Google is trying to put an infrastructure for artificial intelligence agents. Google announced on Tuesday that the SDK for Gemini models will now Supports MCP from Antarbur, MCP, An increasingly common standard for communicating agents to online data sources.

Google is not alone in this transformation. At a different technical conference this week, Microsoft Cto Kevin Scott has set his own vision for “Web Open Agentic”, where agents take action online users. Scott pointed out that the main feature that makes this possible is the plumbing that connects these agents to each other and data sources – i.e. Google Agent2agent and MCP from man.

Despite the enthusiasm, as Bin Tumson notes in the class layer, The agent web has its problems. For example, Thompson notes that if Google sends artificial intelligence agents to web sites instead of people, this largely breaks the form of advertising for the Internet.

The effects can vary across industries. Agents may not be a problem for companies that sell goods or services on the Internet, such as Doordash or Ticketmaster – in fact, These companies adopt agents as a new platform To reach customers. However, it is not possible to say the same for publishers, who are now fighting with artificial intelligence agents of the eyes.

During I/O, the Google Communications Commander told me that “human interest is a truly limited supplier,” and the company aims to launch artificial intelligence agents to give users more time. It may come out all this, but it seems that Amnesty International Entities Summary is likely to take dollars away from publishers – and perhaps destroy the creation of the content on which these artificial intelligence systems depend.

Moreover, there is a problem remaining with artificial intelligence systems around hallucinations – their tendency to make things and present them as a fact – which has become embarrassingly clear with a Google overview. Speaking on the stage on Tuesday, Deepmind Demis Hassabis has raised concerns about the consistency of artificial intelligence models.

“In a few minutes, you can find some clear defects with (AI Chatbots) – some sports in high school that you don’t solve, and some basic games that you cannot play,” said Hasabis. “It is not very difficult to find these holes in the system. For me, for something called AGI, it should be more consistent in all areas.”

The consequences can be long -term. Hallus can be widely lead to users to be more stable in the information they face on the web. They can plant wrong information among users. Which of the result is not perfect.

Google does not seem to wait for advertising-backed companies or artificial intelligence models to catch up with the company-pushing forward with artificial intelligence agents anyway. Google may have done more than any other web supervision company as we know it. But in what can prove a major turning point, the company’s concept of web appears to be re -directed around artificial intelligence agents.