A Threat or An Assistant? -The Relationship between AI and Human Interpreters

Nowadays, there is a group of people calling themselves ‘AI refugee’, that is to say their jobs are more and more being threatened by artificial intelligence. It is predicted that in less than 5 years, 6% of jobs will be replaced by AI, and in 20 years, 60% of the jobs will be threatened by AI. With the rise of machines, can human still make a living? In terms of the job like interpreting, different people hold different views. For those who are not familiar with what the occupation of interpreting is about, it is a translational activity in which one produces a first and final translation on the basis of a one-time exposure to an expression in a source language. Generally speaking, there are two modes of interpreting: consecutive interpreting (CI), which is done at breaks to the exposure and simultaneous interpreting (SI) which is done at the time of the exposure to the source language.

When speaking of the future of interpreting, some people believe that AI will replace human interpreters with the development of neural machine translation (NMT), a new AI technology that saves information about the meaning of phrases, rather than just direct phrases translations. However, others hold the view that although AI can threat interpreters to some extent, there is still a long way for machines to reach the sophistication of human interpreters. As far as I am concerned, the relationship between machine and human beings is interdependent, namely, human interpreters cannot accomplish their works without the assistance of machines like headset, microphone, laptop etc., whereas machines alone cannot convey complicated cultural connotations, humor, emotions, non-verbal communication etc. As a result, taking a proper attitude towards the future of interpreting is of great significance.

The advances and challenges of AI in interpreting

As the world’s most popular translation software, Google translate supports over 100 languages and serves over 500 million people every day by using AI technology. In September 2016, Google Translate switched from Phrase-Based Machine Translation (PBMT) to Google Neural Machine Translation (GNMT), which translates ‘whole sentence at a time, rather than just piece by piece’. There is even an algorithm called Bilingual Evaluation Understudy (BLEU) aiming at evaluating the quality of text. ‘The closer a machine translation is to a professional human translation, the better it is.’

Except from Google Translate, there are a few other machine interpreting solutions in the market including a telephone-based service launched by Israeli startup Lexifone in 2013, the Nara Institute of Science and Technology’s translation app VoiceTra, Clik earbuds deployed by UK-based startup Mymanu etc. All of these efforts prove to be of great promotion on the previous way of interpreting. People nowadays can have the access to understand foreign languages while traveling, working and communicating only through an application on their smartphones. Furthermore, researchers from the Nara Institute are now understood to be working on a lag-free interpreting system for the 2020 Tokyo Olympics, which will reportedly transpose the games’ Japanese commentary in real-time.

Jonathan Rechtman, a Chinese-English conference interpreter with more than 10 years’ experience believes that AI is probably all upside in 3, 10, or 15 years. He also argues that there’s going to be tremendous increase in efficiency and a lot of value that will be unlocked. However, he points out that the rise of machines and the ultra-rich that capture the wealth created by those machines are threatening and dominating the masses of knowledge workers.

This is an extract of a debate between Jonathan Rectman and Hu Yu, CEO of one of China’s most well-established AI companies iFlyTec. The topic is about AI’s impact on interpreting. In the video, Jonathan stressed that we are standing at a turning point in history and the relationship between human and machines is about to be upside down. He also mentioned that AI is not only challenging our job, but also our dignity.

This image has an empty alt attribute; its file name is image.png

The picture above was released by New Yorker in October, 2017 shows that they believe robots are the new white-collar workers of the future while humans are left as pathetic beggars on the street.

Ofer Shoshan is the CEO of One Hour Translation. In his point of view, neural machine technology will carry out more than 50% of the work handled by the $40 billion market within one to three years. His optimistic ideas about neural machines are as follows:

“Today with neural machines, for a growing amount of material and categories, they only need to make a very small number of changes to what a machine outputs, in order to get a human-quality translation.”

Furthermore, Shoshan acknowledges that today on average 10% of a machine-translated document needs to be fine-tuned by humans to meet the standards expected by his company’s Fortune 500 clients. Just two years ago, that figure was around 80%. He also assumed that if machines can do what you can do, then you have a problem. A lot of translators and agencies will tell you that there are certain highly specialized translation services which will require a human touch for the foreseeable future – and that may be true. But for the bulk, 80% of the material that corporate customers pay to have translated on the market today, it will be machine translatable in the next one to three years. And he even stressed that “And importantly, we’re not talking about five to ten years; we’re talking one to three years.”

From what has been discussed above, we can see that people from different trades are all concerned about AI and automation. With the ability of deep learning, some repetitive tasks and even expert jobs are being challenged. AI is working 24/7 sucking up information on global trend from all sorts of sources including social media, financial data etc. The advances of technology did give us a shock but at the same time, the challenges, glitches and hilarious moments AI faced in the field of interpreting should not be neglected.

This picture shows that Google Translate started offering biblical prophesies in exchange for junk input, experts attributed the errors to neural networks’ preference for fluency over accuracy.

There are also other hilarious and humiliating examples. Giving a speech in Beijing earlier this year, hedge fund guru Ray Dalio reflected on his mis-forecasts as a young trader.

“How arrogant!” he thundered to the crowd. “How could I be so arrogant?” But the real-time subtitling program valiantly struggled to render his rhetorical device. “How?” the subtitles asked. “Aragon, I looked at myself and i”.

To conclude, in order to meet the basic requirements of real-time communication without hilarious mistakes, there is still a long way for the AI to truly be qualified as a bridge between two languages. In the translation industry, AI now is equipped with the function of Neural Translation which increases the accuracy and quality of a machine’s output. Teaching machines to truly understand natural language has been one of the biggest challenges facing computer scientists working to advance artificial intelligence for decades. Moreover, interpreting can also be regarded as a cultural exchanges between two languages. The huge differences among different languages can also obstruct the convey of meanings.

The language and culture barriers in interpreting

Today there are between 6000 and 7000 languages in the world, of which about 1000 have some economic significance. That is to say, the technology would need to develop all these languages in order for translation technology take over humans. We shouldn’t neglect that Google Translate supports only about 80 languages, there is still a very long road ahead. Translation is not only about understanding the meaning behind each word, but also the interpretation of context and culture, which captures subtle connotations and nuances in the target language.

From linguistic and cross-cultural perspective, human interpreters must pay much attention to language and expression differences, poetic elements, cultural and custom differences, context, verbal and non-verbal communication etc. when they are dealing with the interpretation work from source language to target language. These are usually not an easy work to do for some compound bilingual individuals who learn the languages in the same environment and context and often use them concurrently or even interchangeably. These situations are happened when a child is raised by bilingual parents and both languages are used in the home so that the two languages are not separate and can be switched between at will, even while speaking. Even though bilingual individuals already have skills for two languages, they are not qualified enough to be an interpreter before professional training. And the United Nations will take fully consideration of the actual language skills before recruiting conference interpreters.

Nowadays, there are more and more critical voices about applications and services for their inability to convey the meaning accurately. As for human interpreters, they usually use context to obtain the meaning of words and take the non-verbal elements into consideration. Due to the complexity of words and meanings in different context, figurative and metaphorical translations are accurate from time to time. This human instinct cannot be easily imitated by the machine through an advanced algorithm.

In a word, AI is threatening most of the low-end repetitive work, and have large probability to replace the people who doing these jobs. However, the relationship between AI and interpreting will not change too much in the short and medium term in the future. Namely, before AI technology is mature and smart enough to notice the subtle differences among languages and can be able to break down the cultural barriers without making hilarious mistakes, we translators and interpreters are safe. Technology serves as the tool and assistant for human interpreters while huge amount of resources can be collected and analyzed and deep-learned by the machine. The balance of this relationship should be kept till the emerging and breakthrough of a more powerful technological transform in the long term.













仲伟合 王斌华. 基础口译[M]. 北京:外语教学与研究出版社,2012.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s