Machines are getting much better at the jobs we humans normally do.
As any translator will tell you, transforming the words from one language into another is a task that draws on an enormous reservoir of training, experience, and art.
It’s something I’ve learned first-hand, too, as I occasionally need to translate news articles and other pieces of communication from Chinese into English. I’m able to call upon my more than three decades of studying and using the language, a language I rely on to do my job, communicate with family and friends, and conduct the most essential daily tasks I rely on to keep my life together.
My goal every time I attempt to complete a translation is ambitious: To make the final text in English sound completely natural, as if it were written in English and not translated from another language. And, importantly, while hewing as closely as possible to the original meaning of the untranslated text.
It’s not an easy task, and the finished, translated product is inevitably just an approximation of the meaning embedded in the original text. That’s why I was intrigued by Microsoft’s recent announcement about what they say is a major new breakthrough in Chinese to English translation.
According to an article published on Microsoft’s AI blog, a team of Microsoft researchers said that they believe they have created the first machine translation system that can translate sentences of news articles from Chinese to English “with the same quality and accuracy as a person.”
Xuedong Huang, the engineer in charge of Microsoft’s speech, natural language, and machine translation efforts, called it a “major milestone in one of the most challenging natural language processing tasks.”
“Hitting human parity in a machine translation task is a dream that all of us have had,” he said. “We just didn’t realize we’d be able to hit it so soon.”
Microsoft has setup a simple test site where you can see the difference between their standard translation engine and their AI-powered one. You don’t even need to understand written Chinese to see the difference in quality of translation, either. Just hit the “Translate & Compare!” button and the system will translate a pre-populated snippet of Chinese text from a news article.
I tested out the new engine on numerous news samples. So what is my assessment? The new engine works really well. I was very impressed with the grammatical accuracy, syntax, and much more natural style of the translations produced by Microsoft’s new AI-powered engine.
Is it perfect? No – not yet, at least. But it’s strong enough to make translators who earn their living from translating Chinese to English very, very worried.
Ming Zhou, assistant managing director of Microsoft Research Asia and head of a natural language processing group that worked on the project, said that the team was “thrilled to achieve the human parity milestone on the dataset.” But he cautioned that there are still many challenges ahead, such as testing the system on real-time news stories.
The implications of this breakthrough are enormous. With over 1.2 billion speakers, Chinese is considered the most widely spoken language in the world. Improving the accuracy and quality of translations from Chinese to English, as this technology seems to do, gives billions of non-Chinese-speakers greater access to the enormous quantities of news, information, and insights generated in this language each day.
The Microsoft team isn’t stopping here, either. Zhou said he expects the methods and techniques they developed to translate from Chinese to English to be useful for improving machine translation in other languages and situations as well. He said they also could be used to make other AI breakthroughs beyond translation.
“This is an area where machine translation research can apply to the whole field of AI research,” he said.