Through the development of artificial intelligence, Google announces that he managed to create an own language.
Google Neural Machine Translation is the recognized system which invented Google for translations and by which has managed to increasingly more accurate translations.
Thus, according to the company, using Dee p Learning Google has been able to create using in artificial teligencia a true machine able to perform translations accurately. Increasing the quality of the content provided and consolidating itself once again as the number one. This intelligent system is used and is developed in the context of the application Google Translate that as we all know.
But where is the innovation?
Recently the application developers have unveiled that the system was created so that initially it will work with the Chinese and English pair. Currently they are surprised by the progress achieved, explaining that through a kind of artificial intelligence system has developed knowledge of 103 languages with different linguistic interpretations.
The creators have mentioned that the system which had been submitted September 2016, to date has been in 103 languages and also analyzes and interpreted in its context, taking complete phrases, which shows greater precision than translations Word-by-Word.
The amazing thing is you artificial intelligence system is that it developed "on their own" in a sense of interpretation and connection of meanings for which had not been created originally.
In this sense, its creators have expressed that the system which, for example, had been developed with a capacity for the Portuguese translations to English and English to the Spanish. Soon he has shown ability to perform detailed translations of the Portuguese to the Spanish, something for which there was not programmed. The latter means that through a system of meanings development itself that new capacity.
To formalize information, Google announced in a statement that: "this is the first demonstration of a true translation, multilanguage not requiring prior exposure (zero-shot)"
The protagonists of the project researchers, announced with clear signs of enthusiasm against the surprising discovery that "the results show that these models learn a form of representation interlingua, for MultiLanguage model, which involves all pairs
The voice and the image replaced text in communication"
The assertion corresponds to the words of Corina Cortés, who is the responsible and benchmark maximum at Google Research based in New York
He is the work area where the magic happens and researchers at the hands of engineers give life and develop the most amazing and innovative projects.
Those who are later applied and are translated through the products offered by the company and will be massively used by users.
More than one hundred researchers and developers teams working intensely, in the production of new knowledge related to algorithms, natural language, safety, artificial intelligence and systems that learn alone and of data processing. Among the many investigations that currently developed to work on generating tools alternative search and communication as it is projected in the coming years will be superadores of the text.
This does not mean that the text and text processors, as neither the search engines will cease to exist, but on the other hand the company is already working on a future need. If you look at young people communicate by voice and images, today text in communication of the
new generations remains in the background.
With this in mind Google is already designing tools to respond to the needs and concerns of the new generation tools that replace the textual language and will focus on voice and images.
Many of the investigations that are underway at Google Research and then deployed in Google products, born of the data that users provide. It is a great challenge for developers know use those data and tracks that users provide through the use, behavior, customs, key words.
Unravel the mystery that holds the interaction to improve the communication tools is the challenge.