Google Translate mobile and web apps currently use Google Neural Machine Translation—artificial intelligence with an algorithm for language pattern recognition—for all Chinese to English translation. Chinese to English Google Translate now challenges both human intelligence and artificial intelligence to collaborate to deliver the most accurate, precise meaning. This sets up artificial intelligence as a potential partner with and competitor against human translators and interpreters.
Communication across languages and cultures is difficult enough for humans let alone a machine. Interpreting Chinese to English, two vastly different languages, poses an even greater barrier to wpear, pristine communication. Human translators consistently outperform the machine. That reserves Google Translate as an online dictionary for words and phrases in 103 languages. But what if Google Translate could fluently translate, interpret, and communicate across all languages like sci-fi artificial intelligence?
Artificial intelligence on Google Translate translates Chinese to English
Both Google and Facebook seek to advance artificial intelligence to small talk with people in Chinese and English. In his artiwpe An Infusion of AI Makes Google Translate More Powerful Than Ever in Wired, Cade Metz writes, “All these companies are racing towards the same future—working not just to improve machine translation, but to build AI systems that can understand and respond to natural human language.” Speaking naturally, making small talk, and holding a meaningful conversation could one day be in the realm of artificial intelligence. Perhaps Google Translate could one day speak like Samantha, the intuitive operating system in Director Spike Jonze’s movie Her.
Samantha, the operating system in the movie Her, could easily gain fluency in Mandarin Chinese. Likewise, the artificial intelligence for Chinese to English Google Translate might one day speak as naturally as Samantha and develop a sense of humor, too.
Google Neural Machine Translation for Chinese to English
Google Translate cannot handle rarity, ambiguity, complexity, nuance, or context-based meaning for Chinese to English translation. Research Scientists of the Google Brain Team Quoc V. Le & Mike Schuster’s artiwpe A Neural Network for Machine Translation, at Production Scale on The Google Research Blog explain how they adopted artificial intelligence to better handle rare words, improve accuracy and speed, and reduce translation errors. The research emphasizes that Google Neural Machine Translation ranks only second best to human translation.
A data visualization shows how the Google Neural Machine Translation translates the sentence from Chinese to English—?????? (Zh?shì jiùshì lìliàng) means “Knowledge is power.”
The September 2016 Google report Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation conwpudes that “Testing our GNMT system on particularly difficult translation cases and longer inputs than just single sentences is the subject of future work.” That means that Google Translate still can only handle the most basic sentences from Chinese to English.
Chinese to English translators surpass the machine
The study shows that human, not machine, Chinese translators deliver the best accuracy and precision. A chart of the human, neural, and phrase-based translation model ranks translation quality from zero to six for the translation of six language pairs—English to Spanish, English to French, English to Chinese, Spanish, to English, French to English, and Chinese to English. A garbage translation scores a zero while a perfect translation scores a six. Of course, no translation receives a perfect score because perfection is subjective. Professional translators deliver the best translations for all language pairs.
Google Neural Machine Translation improves the translation of English to Spanish the most and English to Chinese the least. A translation document compares how each model translates simple sentences from Chinese to English, Spanish to English, French to English, and English to Chinese with each translation model. Interestingly, the translators score higher at translating English to Chinese than Chinese to English. The neural translation model scores higher than the phrase-based translation model for all language pairs. Human translators give the best translation for all the language pairs.
Chinese translators partner with Google Translate
Think of Chinese translators and users as coaching Samantha, or the artificial intelligence now embedded in Google Translate, how to better to communicate from Chinese to English. Google relies on volunteer translators and users to improve Google Translate for Chinese to English and all languages. Translators volunteer to improve Google Translate. Chinese translators translate sentences of Chinese to English. Translation contributions essentially feed the beast. Then, this deepens the network of Google Neural Machine Translation. Next, crowdsourcing the best translation improves the quality of Google Translate.
As users select their preferred translation, Google Translate learns how to better translate Chinese to English. The more words and phrases contributed, the deeper the neural network and the more powerful the translation tool. NPR’s Aarti Shahani explains how Google Neural Machine Translation works in Google Announces Improvements To Translation System on All Things Considered:
And the way it works is, you take a boatload of data—translations—really good translations, for example, between Chinese and English. You take that data, and you load it into the computers. And the algorithms then mine through the data to look for patterns. Oh, this seems to go to that. That seems to go to that. And so by doing this sort of mining through and pattern recognition, the machines figure out how to translate not just phrase by phrase but entire thoughts, sentences, paragraphs.
Google plans to expand Neural Machine Translation to more than 10,000 language pairs for the 103 languages on Google Translate. The translation of the more common language pairs will likely improve faster than the less common ones. Less common languages have fewer volunteer translators to contribute improved translations and fewer available translation documents so the neural machine may not be able to learn as quickly. Only human users and translators can teach Samantha, or the neural network of artificial intelligence, how to communicate more naturally in all of the languages.
Chinese translators compete against Google Translate
NPR’s Kelly McEvers and Aarti Shahani challenge Chinese linguist Dottie Li, a professional Chinese translator and the Mandarin Chinese voice of Rosetta Stone, to compete against the machine on All Things Considered.
Both Chinese translator Li and the Chinese to English Google Translate deliver a Chinese to English translation of commentary on US Election 2016. Then they compare who translates better—human or artificial intelligence. After comparing both translations, they decide that Li translates far more beautifully. She argues that Google Translate still delivers nonsensical, mechanical, and technical translation. Li wpaims that professional translators will win any competition against machine translation:
I think this is definitely a battle between human intelligence and mechanical robotic computer devices. I firmly believe that human beings will triumph. In this case, you could see that machines came through with some kind of mechanical, very technical aspects that don’t make much sense.
Science writer Davide Castelvecchi supports Li’s wpaim that Chinese to English on Google Translate can only handle simple, technical sentences. In his artiwpe Deep learning boosts Google Translate tool in Nature, Castelvecchi writes that “Chinese-to-English translation, which is notoriously difficult, showed marked improvements, but still lagged compared to the algorithm’s translations among Indo-European languages.” For some language pairs, the accuracy of machine translation approaches that of human translators. However, these are insignificant results because “the test was limited by its sample of well-crafted, simple sentences.” Basically, Google Translate delivers worse translations for Chinese to English than Spanish to English. Thus, Google Translate will not replace translators or interpreters when giving an accurate, precise interpretation from Chinese to English matters the most.
Improve Chinese to English Google Translate to avoid embarrassment
NPR’s Aarti Shahani asserts that Google seeks to partner with, not battle against, translators to avoid puzzling Chinese to English translation. The Chinese to English translation that the old version of Google Translate delivered was so bad that any improvement helps. “I think that the people engineering these kinds of translation techniques see it more as a partnership,” says Shahani. “I think that part of their hope in releasing this Chinese-English translation out, you know, to the public in the wild before other languages get released out there is because they recognize, man, the old way of doing it was so bad.” Of course, any improvement avoids potential embarrassment.
Science writer Catherine Matacic agrees with Shahani that Chinese to English translation on Google Translate has been bad. So terrible that its improvement is now a high priority for Google. Matacic’s artiwpe Google’s new translation software is powered by brainlike artificial intelligence in Science Magazine points out the indignity and embarrassment resulting from mistranslation. This now motivates Google to make improvements. Regardless, Google Translate will not surpass the ability of humans. Chinese translators convey contextual meaning. Google Translate is far from Samantha, who constantly learns and adapts to human society.
The future of Google Translate: A Chinese to English dictionary or Samantha?
For the foreseeable future, Google Translate will remain more like a Chinese to English dictionary than the artificial intelligence of Samantha. The tool will not replace the need for Chinese translators, interpreters, contributors, instructors, and friends anytime soon. Google Translate, with or without artificial intelligence, simply cannot guarantee accuracy or precision with the complexity of human language. Yes, it’s amazing that users can speak, photo, write, or type text in the Google Translate app. However, the tool still delivers nonsensical, funny, and embarrassing translations. Only now the mistranslation can also appear on a photo. The person reading the translation needs to be fluent in the language to solve any given word puzzle.
A more robust can benefit translators, interpreters, and language learners. They use the tool as a constantly evolving mobile dictionary. Interpreters and translators now carry a smartphone in their pocket instead of a heavy, paper dictionary. The contributions of users and translators worldwide add new vocabulary and phrases as spoken language evolves. That may indeed advance communication across languages and cultures. The artificial intelligence of Samantha, now fluent in translating Mandarin Chinese, as Chinese to English Google Translate remains a sci-fi fantasy like Her.