Google wants to know Gemini better

In AI Chatbot wars, Google believes that the key to retaining users is to provide content that they cannot get elsewhere, such as the answers that were formed by their online habits.

On Thursday, the company announced Gemini with customization, which is a new “experimental possibility” for its applications in Gemini Chatbot that allows Gemini a fee on other Google applications and services to provide custom responses. GEMINI, with customization, can press the user’s activities and preferences via the Google Environmental System to provide Information designed answers, according to the director of Gemini Dave Citron.

“All of these updates are designed to make Gemini feel a tool that is less similar to the natural extension of you, and you expect your needs with the help of a really personality,” Citron wrote in a blog post presented to Techcrunch. “The first laboratories have found Gemini with allocation useful in brainstorming and obtaining personal recommendations.”

Gemini with customization, which will integrate with Google Search before expanding to additional Google services such as Google and YouTube in the coming months, reaches Chatbot manufacturers including Openai attempt to distinguish between their virtual assistants with unique and attractive functions. Openai recently launched the Chatgpt capacity on MacOS to edit the software instructions directly in the supported applications, while Amazon is preparing to release Alexa “Alexa”.

Citron said that Gemini with allocation is supported by the Google Gemini 2.0 Flash experimental model, which can determine whether it is likely to “enhance” personal data from Google, such as the history of the user search, an answer. Narrow questions informed by likes and hate, such as “Where should I go on this summer?” And “What do you suggest to learn as a new hobby?” , He will benefit more than others, follow Citron.

“For example, you can ask Gemini restaurant recommendations and will refer to the recent foodstuff searches,” or request a travel advice and Gemini will respond based on the destinations that you previously searched. “

If all this looks like a nightmare, it may be so. It is not difficult to imagine a scenario in which Gemini provokes sensitive information for someone unintentionally.

Perhaps this is why Google is doing Gemini with customization-excluding users under the age of 18. Gemini will order permission before contacting the Google search record and other applications, Citron said, and to show data sources that have been used to customize robot responses.

“When you use the customization experience, GEINI displays a clear sign with a link to easily separate the search date,” said Citron. “Gemini will access your search record only when choosing Gemini with customization, and when you give GEMINI permission to call your search record, and when you have a web and application activity.”

GEMINI will be launched with customization to GEMINI users on the web (except for Google and Google’s work space for educational customers) starting on Thursday in the drop -down menu in the application form and “gradually” comes to the mobile phone after that. Citron said it will be available in more than 40 languages ​​in the majority of countries, with the exception of the European Economic Zone, Switzerland and the United Kingdom

Citron indicated that the feature may not be free forever.

“The limits of use in the future may apply,” written in the blog post. “We will continue to collect the user’s notes on the most useful applications for this possibility.”

New models, connectors and more

As additional incentives to adhere to the Grostini, Google has announced updated models, search capabilities, and applications of the platform.

Jemini Advanced, an excellent Google subscription in Google, which is $ 20 a month, can use an independent version of 2.0 flash thinking that supports file attachments; Integration with applications such as Google Calendar, notes and tasks; And the window of the context of 1 million. The “context window” indicates a text message that the model can take into account at any specific time – one million symbols equivalent to about 750,000 words.

Google said that this latest version of 2.0 Flash Thinking is faster and more efficient than the replacement model, and it can deal better with claims that involve multiple applications, such as “searching for an easy -to -you cookie recipe on YouTube, adding ingredients to my shopping menu, finding grocery stores that are still open nearby.”

Perhaps in response to pressure from Openai and its newly launched tools for in -depth research, Google also enhances deep research, which is the GIMINI feature that searches the web to collect reports on a topic. Google said that deep research now reveals “thinking” steps and uses 2.0 experimental thinking as a virtual model, which should lead to “high -quality” reports that are more detailed and “insightful”.

Deep research is now free in an attempt to all Gemini users, and Google has increased the limits of use for applicant clients.

Free Gemini users also get GEMS, which are customizable chat groups that focus on Google themes inside Gemini, which had previously requested an advanced subscription to Guemi. In the coming weeks, all Gemini users will be able to interact with Google photos, for example, searching for images from a modern trip.