Google offers “artificial intelligence mode” in “re -visualization of research”

Digest opened free editor

Rola Khaleda, FT editor, chooses her favorite stories in this weekly newsletter.

Google plans to fix its search engine to add features that allow it to work similar to AI Chatbot, as the company is racing to compete with competitors such as Openai.

From Tuesday, all users will be able to activate “Artificial Intelligence mode” In Google Search and Chrome Browser, which will provide a conversation, question and answer closer to Chatgpt than Openai, instead of the traditional link menu.

Soundar Pachai, CEO of Google Parent Alphabet, described the move as a “complete re -conception” during the annual developer I/O conference at Mountain View this week.

The version represents the latest effort to pump AI in the Google dominant search engine, even when investors worry that new products will confirm their cash cow.

It follows the new “artificial intelligence situation” introduction From “Amnesty International Overview” at I/O last year. The brief answers created from the artificial intelligence of common topics became common for users, but they reduced the clicking rates on the ads that the company relies on in many of its revenues.

Pichai was criticized because of the slow AI versions after Google was previously a pioneer in the industry that developed the research behind the artificial intelligence boom. This allowed startups such as Openai and Anthropor to grow into billions of dollars in companies that are starting to challenge Google’s monopoly on online inquiries.

In response, Google has been forced acceleration The timeline of its launch and made quick improvements on the Great Gemini language model. It claims that the latest 2.5 version exceeds the Chatgpt from Openai and Claud in many standard tests, especially coding.

Beshai said that Google still has a wide distribution advantage to competitors with 8.5 billion daily inquiries. “The research brings Amnesty International to more people than any other product in the world,” he said.

Executive officials were tight on how they planned to integrate ads into search offers, browsers and admirers based on the application. Google achieved $ 50 billion in advertising revenues in The first quarterMore than half a total of 90 billion dollars for the alphabet in general.

Since it flying to artificial intelligence, Google moves further from its free advertising model. The cost of the “AI Pro” standard subscription plan is $ 25 per month.

Those who want early access to the most advanced features will have to pay $ 250 a month for the new “ULTRA” package, which is more expensive than the top $ 200 in Openai per month.

Google also joined companies’ rush Presenting artificial intelligence agentsWith the initial model called Project Mariner to be offered to subscribers during the summer. Users can allow the agent to control their browsers and other programs to carry out tasks such as booking trips, shopping and collecting complex research reports.

Google users can also grant Gemini access to their historical data to provide more “personal context” of procedures and responses. However, experts raised Fears On privacy where artificial intelligence companies begin to build long memories in their models.

“Our recent updates to the Grostini are important steps towards opening our vision of global artificial intelligence,” said Sir Dimis Hasabis, head of the artificial intelligence research department. “He is smart and understands the context you face, and he can plan and take action on your behalf via any device. This is our final goal.”

Many of the “Agentic” features that were announced on Tuesday are already offered by smaller competitors, as Openai launched its own versions of computer use agents in October and January, respectively. However, these products remain great and often struggle with basic tasks. Google can acquire an advantage if it is able to provide a more reliable experience using its wide resources and resources from user data.

Love MicrosoftThe host of its developers in Seattle this week, Google said it will adopt the protocol of the Hothropic (MCP) model protocol, a frame that allows the artificial intelligence agent to communicate with each other in a unified coordination, necessary to move between multiple applications and web sites.

Google has also inspected an upgraded version of Project ASTRA, a “multimedia” agent who can respond and work in the actual time and vocal orders, using a mobile phone camera or smart glasses to look at the world. A man’s video show showed advice on how to fix a bike, and ask for spare parts from a local store while working.

Google ultimately plans to integrate ASTRA directly to search, but it will also allow access to devices that operate the Apple IOS program, and not only those who use Android Mobile operating system.