Will AI simultaneous interpretation tools replace human translator? We believe NOT. Setting aside the complexity of the language itself, we can take a closer look at the actual applications of simultaneous translation. In many occasions where people from different language backgrounds need to exchange ideas, simultaneous interpretation is engaged to streamline the whole process of communication. AI simultaneous translation tools, if deployed, should not only learn to translate, but also learn how to chat. And there’s a lot of room for their improvement here.
So what will be the best scenarios for these clever robots to give their play?
- Shared AI Simultaneous Interpretation Devices, A Market for the General Public
When people travel abroad, they often encounter language communication problems. However, not everyone can afford a professional oral interpreter. Now, if a wearable device or a mobile phone can interpret for you simultaneously, it will surely reduce the cost of going overseas for many people. Isn’t it cool to have your own “simultaneous interpreter” travelling with you?
Smart hardware has always been a hot spot for AI. Just last year, Microsoft partnered with huawei to embed Microsoft’s neural network machine translation in the Mate 10, arguably the first example of a neural network machine translation running on a terminal. If the hardware equipment of AI simultaneous translation is born, the more likely business model is to rent or share, that is, rent according to the demand, and there is a specialized technology company responsible for leasing, just like sharing the bicycle, our usage fee will be reduced to extremely low.
However, the application scenarios of such AI are not in the complex conference site, but in daily life, travel, etc., and the construction of corpus will be more simple. In this way, AI co-communication is only more beneficial to the people, but it will not replace the professional human co-communication in a special situation, such as financial conference, medical conference, etc.
By combining the simultaneous translation of AI with hardware devices, a usable voice interface can be created, and users’ experience in mobile terminals, wearable devices, smart homes, smart cars and other smart devices can be improved to a large extent, so as to truly realize the human-machine integration in the intelligent era at the interactive level.
- AI Simultaneous Interpreter Acting As the Examiner of Human Linguists
The demand for simultaneous interpretation has multiplied in today’s market, but the number of qualified simultaneous interpreters has grown very slowly. According to some industry study, it appears not easy to get a dozen of top-elite human simultaneous interpreters who could consistently score over 90% on the language accuracy for any half-hour conference interpretation services.
At the other hand, linguists who have successfully obtained advanced Chinese to English translation certificates may not necessarily be qualified for real-world simultaneous translations, for the latter would require very specific skills and long-time specialized trainings. Some qualified simultaneous translators do not necessarily have interpretation certificates.
At present, China does not have an official institution to take charge of the relevant matters of simultaneous translation, nor does it have a set of unified standards to evaluate the work of simultaneous translation. In the face of such a dilemma, we may be able to use some imagination in the AI simultaneous upload. People can take advantage of the characteristics of AI digitalization and standardization, and train AI into a single functional assessment machine based on the database. In view of different application scenarios, simultaneous interpreters can be evaluated and rated, so as to standardize the talent market.
Let’s take a look of the driving training robots: driving training robot includes high-precision GPS navigation technology, inertial technology and virtual sensing technology, video detection, data processing, wireless translation, fingerprint identification and other high-tech, can accurately record and judge the driver’s real ability to operate a motor vehicle.
Similarly, AI simultaneous translation can also observe and judge candidates’ translation ability, as well as their familiarity and understanding of translation rules in various scenarios. This system can reduce the labor intensity and human factors of the examiners, and ensure the fairness and impartiality of the examination, as well as scientific and accurate assessment methods.
To put it in simpler words, our goal is to automate assessment, select or train real talents through intelligent machines, rather than replace human translation ability. What’s more interesting is that in the assessment process, AI can constantly absorb new nutrients and increase the corpus of spoken texts – so why not.