Artificial intelligence has been at the heart of Google products for several years. The deep learning algorithms they index information on the search engine, have made possible the enormous progress made since 2016 by Translate (the translation software) and allow the Nest home automation system to learn our habits.
However, thanks to constant progress in the field of artificial intelligence, the tools that make use of it continue to expand their functions. Some of these innovations were presented by Google during the I / O conference preview (here the hardware news)starting with those relating to the search engine: “If we think about how people have traditionally searched for information, talking to each other or looking around, we realize that performing this operation only by typing in a specific space is quite limiting – it has explained Nick Bell, senior director di Google Search – We want people to be able to find the information they are looking for everywhere and in the most intuitive ways possible ”.
Smartphone e smartwatch
Google I / O 2022: from the Pixel 6A to Watch and Buds, the new hardware
by Giuditta Mosca
12 Maggio 2022
The news of the search engine
A step that has already taken place in this direction is obviously that of Lens, the platform that allows you to search for information simply by framing an object and exploiting for this purpose, among others, the object recognition algorithms: “Today 8 billion searches per month with Lenswhich has grown 3 times from a year ago, ”Bell added.
These tools are now gaining new potential thanks to what Google calls Multisearch: the possibility of cross the visual research with the textual one, for example photographing a pair of red shoes, but typing the word “green” in the search engine to ensure that those of your favorite color are shown. That’s not all, because this same function (which will be introduced in English with the aim of expanding it to other languages) will also allow look for which shops nearby have what you are looking for.
“Let’s imagine taking a photo of a broken part of the faucet – clarified Bell – With this tool it will be possible to search for it even if we don’t even know its name and above all it will be possible to immediately find information on which shop in the area has the piece we need in its catalog. “. It will also be possible photograph a dish seen on television to search which restaurants you can eat and more: “This is possible because the system compares the image taken with everything on the Web and obviously also using the contributions uploaded directly by users”.
In addition to artificial intelligence, even an innovation such as augmented reality is at the center of some of the innovations presented by Google: Scene Exploration is another search tool that allows, for example, to frame the shelf of a library and receive information on the books present, also exploiting for this purpose various artificial intelligence systems such as there computer vision or natural language processing.
transport
7 Google Maps functions that have changed the way we travel
by Emanuele Capone
09 June 2021
The news of Google Maps
However, augmented reality is also at the center of some news regarding Google Maps. Let’s say you have to find your seat in the stadium – in the coming months it will be possible frame what is in front of you directly to you to view arrows through your smartphone that (digitally superimposed on the physical environment around you) will lead you to the right place.
The same tool, called Global Localization, allows you to view indications that guide us to the nearest stall in which to leave the bicycle or easily find parkingbut it can also have playful applications: during I / O, for example, it was shown how to visualize in augmented reality a group of dragons appearing in front of the historical monuments of Tokyo.
Another novelty concerns the Immersive View of Google Maps, which will be initially available in New York, Los Angeles, San Francisco, London and Tokyo (in the course of 2022) and thanks to which it will be possible to explore a place even before reaching it: showing not only the cities in a subjective view, but also from a bird’s eye view in 3D or taking us directly within places (museums, restaurants, shops) that you want to visit.
More intuitive assistant
Other innovations based on artificial intelligence concern the communication with Google Assistant and Nest Hub, the smart speaker with display through which to manage the home automation at home and beyond. To communicate with Nest it will now be sufficient to look at it, without the need to say the command “Ok Google” (a novelty called Look and Talk). Taking advantage of the image recognition algorithms, Nest itself will realize that we are talking to him (for reasons of privacyGoogle states that the processing of this data will only take place on the device and will therefore not be transmitted to third parties).
Then there are the so-called Quick Phrases: these are quick phrases (such as asking to set a timer or turn on the lights) that will be recognized by the Assistant’s artificial intelligence system without having to say the aforementioned activation command. New improvements also affect the understanding model of natural languagewhich will allow us to understand when the speaker is correcting themselves and to better manage our breaks and interruptions.
Translate if wide
Finally, Google Translate is about to add to the database 24 new languages spoken around the world by about 300 million people, thus covering 133. For this purpose a new tool (called Zero Resource Translation) has been used which represents an important advance in the field of artificial intelligence: “To train the algorithm, previously, millions of examples of a sentence translated from one language to anotherwhich caused a serious bottleneck – explained Isaac Caswell, Google researcher – Now we are able to create a translation model even for a text of which we only have the source language ”.
This tool, which can potentially extend the use of Translate also to less common languages, has so far been used only on some languages, giving priority to those with a larger population of speakers. For the first time, however, Translate was thus able to support some native languages of the Americas, such as Quechua or Maithili: “For us it truly represents a milestone – concluded Caswell – Above all with a view to increasing inclusiveness “.