VIEWPOINT MAGAZINE Volume 5.4 May 2017 | Page 26

Lifestyle
Life with“ Smart Translators”

Lifestyle

Rebuilding the Babel Tower

25

Life with“ Smart Translators”

W

By James Kwak
hen the Lord saw His people constructing the Tower of Babel, daring to challenge God’ s authority, He confused the people with different languages. Sadly, many students have had to study various languages since then. Subjects such as the SAT, English class, and Indonesian class are burdening the students of Mountainview with complicated languages. Luckily for these pitiful students, scientists have developed an“ ultimate translator.” This translator can instantly and flawlessly translate any sentences among different languages, freeing the burdens of language learning. With this translator, different languages of the world can be unified, just like the people who were constructing the Tower of Babel.
So, what is this“ ultimate translator” and how does it function? The translator is named Neural Machine Translation( NMT). Open the Google Translate in your web browser and try translating anything. There, you just saw neural machine translation happening. In fact, Google is one of the leading pioneers of NMT technology. Unlike its scary name, its fundamental mechanism isn’ t as complicated. Scientists programmed an artificial network of neurons, similar to how our brains are shaped. The neurons act just like real neurons. When new data are given, the neurons fire and send out electric signals. The signals travel through the neurons, causing them to be activated. However, as the signals pass through the neurons, the signals slowly diminish to the point where the signals may not activate the neurons. This activate-diminish cycle is repeated numerous times with already translated data to“ train” the network. As a result, when new information is given, they are translated through the established system. Sounds confusing and unrealistic? Science is developing as rapidly as ever. Some scientists are even predicting the removal of language barriers in five years. After all, a unified, integrated world might come in near future.
However, we all know that Google Translate rarely produces decent translations. There are two primary reasons why it is not doing a good job. First, NMT is heavily based on accumulated data. If there are not enough existing translations or if they does not have consistent quality, the neural network of Google Translate will not be trained properly. Second, NMT is a developing technology which is recently adapted to the Google Translator. NMT for Korean-English translation was first started last year, 2016. Scientists are still finding ways to adapt and improve NMT. Sadly, days are still far for the perfect translation of Google Translator.
Try imagining how the world would look when the translator enters into our daily life. Korean students may be searching German internet websites for their homework. Seniors could easily apply to foreign colleges without having to learn new languages. Non-native English-speaking students would comfortably communicate with their English-speaking friends. Language barriers that prevented people from communicating with each other would be eliminated through the help of the translator. However, there may also be unpleasing consequences. Racial hatred might spread uncontrollably through the internet with less obstacles to halt them. Competition would be intensified since everyone from different languages would be equally capable of communication. Previously, international students only had to compete with English speakers when they applied to American colleges. Now, students of all languages are capable of competing for college acceptance using the translator. In these ways, this technological development would not only bring delightful benefits but also dreadful social consequences.