Google Translate using Neural Machine Translation to improve ‘more in a single leap’ than last 10 years combined

Google Translate is ten years old and in that time, it’s grown quite a bit and gotten increasingly accurate. However, Google made a giant step in improving translations earlier this year through the use of machine learning. Now that project is expanding…

Google first used machine learning to improve its translations of Mandarin Chinese and the results were nothing short of impressive. The system reduced errors by up to 85%, a massive leap forward in accuracy. Neural Machine Translation for translations is clearly beneficial, and that’s why Google is expanding it to eight additional languages: English, French, German, Spanish, Portuguese, Chinese, Japanese, Korean, and Turkish.

In 10 years, Google Translate has gone from supporting just a few languages to 103, connecting strangers, reaching across language barriers and even helping people find love. At the start, we pioneered large-scale statistical machine translation, which uses statistical models to translate text. Today, we’re introducing the next step in making Google Translate even better: Neural Machine Translation.

Google says that these eight languages will cover over 35% of all translations and include about a third of the world’s population in terms of their native language. Over time, Google also plans to bring Neural Machine Translation to all 103 languages which Translate currently supports.

In Sundar Pichai’s own words:

Using machine learning, #GoogleTranslate is improving more in a single leap than we’ve seen in the last 10 years combined https://t.co/YPS0H7bFxy

— Sundar Pichai (@sundarpichai) November 16, 2016