Google has announced that Google Maps will also feature artificial intelligence capabilities, with large-scale language models (LLM). The feature will allow the platform to analyze registered establishments and “talk” to users to find what they are looking for even if they don’t know very well where they want to go.

The new tool will be released this week, but only in the U.S. and to select participants in the Local Guides program, the community of members who contribute information, reviews, and photos to Google Maps. Feedback from these power users will be used to refine the AI before rolling it out to more users.

According to Google, the more than 250 million registered places and the contributions of the more than 300 million Local Guides will be analyzed by LLMs. The idea is to better understand what each establishment is and show it in search results, even if the terms typed by users are not exactly the same as those in the registered description.

AI Helps Answer Less Specific Questions
In the announcement blog post, Google gives an example: when searching for places with a “vintage vibe” in San Francisco (USA), Maps may show thrift stores and vinyl record stores in the city, with curated photos and AI-generated summaries of reviews.

Then, the user could ask “And for lunch?”; The app would understand it as a continuation of the previous request and would show places with an old face. The company also mentions “rainy day activities” and “options for children” as other possible search sequences.

Google Maps already has a tiny bit of AI
In October, Google Maps received an update that already featured artificial intelligence. The platform started to understand searches using more vague terms, such as “things to do”. In the example given by the company, when searching for this in Tokyo, the app would give themed results, such as anime, art exhibitions, and cherry blossoms.

This Thursday (1st), Google announced several new features involving AI. In addition to the new tools in Maps, the company has released the Gemini Pro model for Bard in more than forty languages, including Portuguese. The wizard will also be able to create images, but this feature is restricted to the English version for now.