The new AI tool designed by Google is meant to improve the Google Translate feature, making possible the translation of unknown languages. Isn’t that weird? How come the technology has advanced so much that it allows the creation of a new secret language? The new internal self-taught linguistic was invented by the Neural Machine Translation (NMT).
In September 2016, Google has announced the launch of their new AI tool primarily designed by Neural Machine Translation. Many people must have wondered how a secret language was even possible to be developed. The NMT can translate a whole sentence, not only separate words or expressions. The fact that an entire sentence can be rewritten in a different language will help the user maintain the meaning of the phrase, without altering the meaning of the complete sentence during the translation process.
This technology is capable of analyzing wider contexts, thus enabling an accurate translation, without you engaging in all sorts of paraphrasing or other editing linguistic processes. The translation will use reliable grammar, the use of the tool being different from the previous uses of Google Translate. The translated text will sound more humane, being fluent and flawless.
The Neural Machine Translation was designed as a system which is based on end-to-end learning, being capable of developing new and even more accurate translations, making them sound in a natural way. In the past, this feature of Google was not as advanced as it is today, being improved over time boosting its capacity from 55% to 85%.
One of the primary goals of Google regarding this feature was to enhance the translation from Chinese to English. The new AI tool can be used for Google Translate for the most popular languages, namely, Turkish, Korean, Japanese, Chinese, Portuguese, Spanish, German, French and English. The NMT developers intend to develop it for all 103 available languages.
The first test regarding the new update was constructed on a Korean-English translation, and also vice versa. Specialists had also checked the Japanese-into-English translation to verify the accuracy. The purpose of the developers was to test if the device can translate from Korean to Japanese without using English as an intermediate. The result was surprising. The tool was able to provide a correct translation.
However, scientists are still triggered by this success, questioning if it’s possible for NMT to produce an internal language, helping it translate by analyzing sentence structures and concepts.
Image courtesy of: wikipedia