The power of digital translation tools like Google Translate is undeniable. Machine learning in translation has evolved impressively since 2016 when the tool first launched. Artificial intelligence sorts through tens of thousands of entries, using them to mimic human patterns. The tech captures everything from common phrases to slang, but anyone who translates on a regular basis has had a moment where the machine spits out the wrong text.
For all its advances, machine-powered translation still isn’t as failsafe as hiring a professional translator. Using machine learning translations in public-facing materials without the finishing touch of a professional translator is tempting, but risky.
The difference between Google Translate and human translation
Human translation involves using a professional translator, who uses their proficiency in two languages, to convert text from one language to the other. Google Translate is a digital tool that uses neural machine translation (NMT), as well as human contributions. NMT is a form of deep machine learning that imitates the way human brains develop speech. Combined with data from human contributions, Google Translate’s NMT technology takes text in one language, and instantly converts the text to a second language of the user’s choosing.
Benefits and pitfalls
Depending on the purpose of the translation, service needs will differ. There are clear benefits to both professional human translation services, as well as using a service like Google Translate, or another AI tool for translation.
When it comes to human translation, the biggest draw is peace of mind, knowing the translation has been completed to the highest degree of accuracy. Human language is complex. The intricacies of effective communication, like tone or humour, are most precisely conveyed by a professional translator. There’s no substitute for people-power in this instance – a professional translator has a full range of lived experience to pour into the choice of words in the output language. Plus, there’s no need to worry about mistranslations or other mistakes in the finished product.
There are two main pain points to human translation services: cost and speed. Because human translation is a manual process, where someone is converting text sentence by sentence, it isn’t instant. For the same reason, professional services come with a cost. Since a human’s time is required to create a finished product that’s high quality and accurate, of course, it’s necessary to set aside a budget to pay for the higher degree of service.
Machine translation is impressive for its speed. Input text is translated nearly instantly, and there is a big range of languages, and varying characters and symbols, that Google Translate is able to process fast. The other obvious benefit is cost – or lack thereof. Google Translate is free to use, and easily accessible at a moment’s notice, as opposed to booking a professional translator. Just because it’s freely available, doesn’t mean it should be used in all instances!
The challenge with machine translation is accuracy. A human translator will still be required if the team doing the translation doesn’t have someone who speaks the output language to proofread the text the machine produces. In moments where the final translation is needed for something like a report or a speech, it’s impossible to be sure the translation is perfect with machines alone.
Machine vs. human translation in the wild
The next logical question is how to decide when to use human translation and when to go with machine translation. For quick, low stakes translations, like a casual conversation between colleagues who speak different languages where you are needing the gist of it, Google Translate is perfect. It’s fast, and even if the finished product isn’t perfect, chances are the translation is accurate enough to be legible by someone who speaks the output language. Human translation is ideal for moments when it’s important to capture things like local language variations (slang), or there’s a need for moderation between group members with different native tongues. In these instances, an interpreter with the ability to localize language choices in real time is invaluable.
Examples of other AI translation tools on the rise
Google Translate is consistently ranked among the top machine learning translation services. Microsoft’s Bing Translator has a similar interface to Google’s, and it’s able to translate 100 languages, with the ability to read text from websites and images, as well as interpret the human voice. The list of machine services is growing all the time, and the more people use them, the smarter they become.
For accuracy and flexibility across the many situations where translation may be needed, there’s no beating the power of a professional, human translator. Still, machines are an excellent stop gap in moments where a translation is needed fast, and accuracy isn’t too much of a concern. Between the two, any professional setting would benefit most from the human touch.
 Source: https://en.wikipedia.org/wiki/Google_Neural_Machine_Translation