Google has recently shown a lot of ambition by releasing its transltion toolkit into 345 different languages. These languages are extremely varied and range from the likes of Abkhazian to Zulu. This is the toolkit that Google uses to improve and ultimately create new language pairs that can be used for automatic translation works. And the best part for Google is that all the work is done by someone else, for free!

Whatever Google is, the company is extremely intelligent in certain cases. The translation toolkit is used numerous academics all over the world to translate languages that have no automatic translators, like Maori. It is also used to translate currently supported Google translate languges like Welsh and that input helps to improve automatic Welsh translation.

Google is extremely good at doing this actually. Each and every second, Google is serving millions of searches and they are actually learning from it each time. Whenever an user clicks a link or modifies a search, the great Google learns from it. It mines data from it.  A recent tour of Google search development labs showed that each search is also likely to be part of a lot of multiple trial runs of codes that engineers use to see if it enhances the accuracy of searches. From learning contextual differences to learning the difference between names and other words, Google is becoming more adept each day at doing things that only human beings have been god at so far. I am not sure if I should be excited or paranoid.

The same things work when it comes to this translation toolkit. It helps translators input the language pair into Google and then reap the benefits themselves. When multiple people use it, they all benefit from the work but the main beneficiary is Google. The translators ensure that the translation data is stored for future use and Google slowly gains enough data to launch a new language pair.

Google currently has 52 pairs of language that it can translate. That is more than any other machine translator there is. While it has accomplished this over the past few years, it continues to get better. Every time someone uses, Google asks for better translations. Every time someone obliges, Google learns from that side by side input. The translator toolkit basically does that same thing at a much lower level on a larger scale.

And the best part is that mostly academics are using the tool because you need to know the rare languages to translate them into another language that you know well. Only academics are capable of that in most cases. It is helping them preserve languages that are on the verge of extinction or are extremely difficult to translate in to a more well-known language.

Google’s motto may be ‘do no evil’ but I sure am worried about Google being able to access so much data and actually process and ultimately understant it. Engineers at Google tell us that at the present moment, the toolkit supplies nothing that can compare to the googles of data that Google can access online. But the data that does come through the toolkit is very precise and useful. Being absolutely side by side translations, they help Google learn new things like no other.

So as Google learns to understand human thoughts that it previously did not understand, let us step back to the time when speaking a non-english European language was enough to take care of eavesdroppers. I say this because Google is ultimately working towards real time voice to voice translations. That will be an absolute godsend for travelers, lovers and a host of other people. But it will also likely find its way into ‘evil’ applications. Sadly, it can’t be helped. Nuclear power is both good and ‘evil’ and so is Google — really good at what it does and hence really ‘evil’.

[read]

People Have Also Been Looking For: