Google Translate update shows the true power of machine learning

The mechanics behind machine learning and artificial intelligence can tax many a non-techy brain at the best of times.

But the travel industry should be aware of how prevalent it will become in various processes (especially in customer service) and how it can also solve many problems.

Here is a good example, and one which is at the heart of one of the fundamental aspects of the travel experience: language.

In late-2016, Google quietly pushed out what some are considering to be one of the most “astonishing” updates in the field of machine learning.

The ten-year-old Google Translate platform, which apparently deciphers 140 billion words every day across 103 languages for thousands of websites and search terms, switched from its previous system to something called Neural Machine Translation (Google NMT).

Specifically, Google created what it calls “zero-shot translation” – essentially, allowing it to automatically transfer the knowledge required to make a translation.

In other words: Google invented its own language to help it translate more effectively.

According to tech consultant Gil Fewster:

“It [GNMT] developed the ability to learn from the people who used it. It learnt how to make educated guesses about the content, tone and meaning of phrases based on the context of other words and phrases around them.”

And this is where the brain really starts hurting a bit.

The nerds at Google wrote:

“Is the system learning a common representation in which sentences with the same meaning are represented in similar ways regardless of language — i.e. an ‘interlingua‘? Using a three-dimensional representation of internal network data, we were able to take a peek into the system as it translates a set of sentences between all possible pairs of the Japanese, Korean, and English languages.”

The result?

“Within a single group, we see a sentence with the same meaning but from three different languages. This means the network must be encoding something about the semantics of the sentence rather than simply memorizing phrase-to-phrase translations.We interpret this as a sign of existence of an interlingua in the network.”

“We interpret this as a sign of existence of an interlingua in the network.”

And there it is, ladies and gents. Computers learning how to do something as freeform and creative as language translation.

NB: The Google blogpost on it is here.

Share on FacebookTweet about this on TwitterShare on LinkedInEmail to someone
Kevin May

About the Writer :: Kevin May

Kevin May was a co-founder and member of the editorial team from September 2009 to June 2017.





Your email address will not be published. Required fields are marked *

  1. Dan O´Beirne

    Wow Neural Machine Translation (Google NMT) sounds revolutionary! Thanks Kevin


Newsletter Subscription

Please subscribe now to Tnooz’s FREE daily newsletter.

This lively package of news and information from Tnooz’s web site provides a convenient digest of what’s happening in technology that drives the global travel, tourism and hospitality market.

  • Cancel