Quality prediction: AI for faster human translation

Translation quality prediction uses Artificial Intelligence to speed up human post-editing and revision. It promises to make translation more efficient and less costly.

Text by Adam Bittlingmayer


Image: © peterschreiber.media/istockphoto.com

The amount of content on the internet is growing exponentially. Yet, internet users can only understand a fraction of internet content because it is not in a language they understand. According to consumer data provider Statistica, more than six billion people in the world cannot understand English, and four billion of them are already online. These people mainly live in societies with the fastest growing populations, economies, and internet adoption.

As a young software engineer working for Android and later Google Translate, I watched how basic access to content grew from reaching millions to billions of people, thanks to affordable devices, internet access, and machine translation. For the vast majority of people, machine translation is the difference between bad translation and no translation. Human translation is just too slow and inefficient at scale. Even post-editing machine ...