Recently, new breakthroughs have been made in related technologies in the field of artificial intelligence, and applications and software related to AI translation have become increasingly abundant. In addition to several popular translation software in recent years, such as Google Translate, new AI tools such as ChatGPT are increasingly being used for language translation, which can even generate more accurate and fluent translated text.
How will AI impact the translation industry? Focusing on related topics, China News Agency's "East-West Question" previously interviewed Liu Heping, honorary dean of the School of Advanced Translation and Translation of Beijing Language and Culture University, in Paris.
The following is a summary of the interview:
China News Agency: We see more and more applications of artificial intelligence in the field of translation, especially the increasing influence of new AI technologies such as ChatGPT. Will AI have a big impact on human translation in the future?
Liu Heping: In my opinion, the development of artificial intelligence can be compared with the development of mobile phones. Mobile phones initially only had a call function; With the development of technology, mobile phones have more and more functions, but mobile phones have developed to the present can not replace people. The same is true for artificial intelligence. ChatGPT may be considered to have had an impact on human translation, but in fact, the technology is providing more help to humans and cannot replace human labor. Humanity has also evolved in the process.
In the field of translation, AI can provide more translation services to humans. Everyone's translation level is different, if a person's translation level in a language is not as good as AI, then his related work will be replaced by AI translation. In the future, the translation industry needs more high-end translations, that is, the translated text translated by humans must translate not only the basic information of the original text, but also the emotions of the original text.
I've done a lot of experiments with translation using ChatGPT. It can translate the original text smoothly and without grammatical errors, but the translation translated by this AI is without emotion. The joy, anger and sorrow I want to express through the original text, the AI translation can't fully express it, at least for now. Human translators should leave a good place for their own development in terms of "innovation" and "emotion".
China News Agency: But there is some concern from all walks of life, that is, will human translation really be replaced by artificial intelligence in the future?
Liu Heping: Now translation students and parents are worried about this, and public opinion is also worried about this, and I personally think that this worry is unnecessary. AI translation can never replace human translation. In our early years, we used manuscript paper to write articles, then computers, and then we started using speech recognition technology. As technology advances, our productivity is increasing, and now people are clearly reluctant to go back to the days of writing articles on paper. Therefore, I suggest that young students should face the future, embrace technology, and fully explore the value of human beings.
In the future, perhaps 90% of the translation may be done by translation software, but the remaining 10% will still need more skilled translators to complete it, and someone will always need to implement this "last mile" translation work. Therefore, this puts forward higher requirements for the cultivation of translation talents. AI technology is changing rapidly, and even though I have retired, I have been learning and exploring in related fields now.
After the experiment, I was thinking about what ChatGPT is not enough to do in translation, and what are my advantages over these shortcomings, so I came to the conclusion that AI translation cannot replace human translation.
China News Agency: For example, in literary translation, will artificial intelligence not be able to replace your work?
Liu Heping: AI can now do some literary translations, but the translations are very bland. The emotions that the author wants to express through literary works, AI cannot translate.
Therefore, I tell students not to be afraid of the development of technology, what needs to be done is to improve personal capabilities, learn to use AI well, and let this technology "plug another pair of wings" for translators, so that they can "fly higher".
China News Agency: In 2018, CIPG cooperated with iFLYTEK to build an artificial intelligence translation platform, and the United States is also actively investing heavily in the research and development of an artificial intelligence translation platform. As far as you know, what is the current level of artificial intelligence translation platforms?
Liu Heping: In the past, I used Google Translate a lot, mainly to translate between Chinese, English, French and other languages, and some foreign software was better translated into European languages. DeepL translation is now used by students more and the quality is good, but how good it is is still debatable. In general, the completion of European language translation of AI translation platforms is relatively high, but the completion of translation between Eastern and Western languages is relatively poor. The basic problems of the translation of these translation platforms are mainly that they are cumbersome, and the connection between sentences and logic are relatively poor. Although there were no major grammatical errors, the overall feel of the translation was rather blunt.
The AI translation platform has a self-learning function, we have done relevant translation experiments with Microsoft before, and enter the speaker's presentation into the platform before the start of the conference, so that the platform can understand the general content to be translated in advance, and the error rate of the AI translation platform for the conference translation will be significantly reduced. This has a certain similarity with human translation. But AI translation platforms can't do completely accurate translations. For multilingual simultaneous interpretation, the AI translation platform cannot handle it.
In July 2019, the iFLYTEK delegation brought artificial intelligence products such as iFLYTEK translation machine and iFLYTEK intelligent office book to the 2019 Australia-China (Melbourne) Expo, showcasing the achievements of iFLYTEK and its ecological cooperative enterprises in the field of artificial intelligence. Photo by Zhang Xiaoli
China News Agency: You presided over the research and development of "Cross-language Terminology Database Construction and Application Development for the Winter Olympics", which takes the 2022 Beijing Winter Olympics and Paralympic Winter Games as an opportunity to have a termbase in eight languages: Chinese, English, French, Japanese, Korean, Russian, Arabic and Spanish. Why create this termbase? Is it meeting expectations? Will it also be applied to other activities?
Liu Heping: The Beijing Winter Olympics Organizing Committee launched the Beijing Winter Olympics Language Action Plan in 2017, and Beijing Language and Culture University set up corresponding projects to serve the Winter Olympics. In 2021, we delivered the third edition of the "Interlingual Glossary for the Winter Olympic Games", which will play a corresponding role during the Beijing 2022 Olympic and Paralympic Winter Games.
The development of a cross-lingual termbase is actually a pioneering work. All linguistic technical information has been lost after each previous Olympic Games, and the Beijing Winter Olympics provided a unique opportunity to develop this cross-lingual termbase. The IOC decided to use these eight languages for the Beijing Winter Olympics, and we set up expert groups in eight languages, which met regularly to study and promote the development of the relevant terminology database, and finally compiled a total of 132,000 terms corresponding to the eight languages. This is the world's first multilingual intelligent platform related to the Winter Olympics, covering the core terminology, operational terminology and common vocabulary of the Winter Olympics, filling the gap of Winter Olympic terminology.
We hope that after fulfilling the task of serving the Beijing Winter Olympics, the termbase can become the linguistic heritage of the international Olympic movement, and the future Winter Olympics can also be used as a reference. I personally hope that in the future, with the support of the state, this termbase can be expanded into a cross-language terminology platform for national special sports, such as football, tennis, marathons, etc. Next, if you want to apply the termbase to other sports events, such as the Hangzhou Asian Games and the Paris Olympics, you should add relevant content. The idea of related platform construction has been very clear, and there is room for further promotion in the future.
China News Agency: Can this termbase be applied to artificial intelligence scenarios?
Liu Heping: Yes, the "cross-language termbase for the Winter Olympics" is not a simple vocabulary platform, but more like an encyclopedic knowledge, which can provide the corresponding context when querying relevant terms. We can understand in what context the relevant vocabulary is used. BLCU and Tsinghua University are also developing other AI-related research projects, and we hope to combine these research projects with termbases to make them more useful. Of course, this also requires continuous investment in the later stage.
Respondent Profile:
Liu Heping, Honorary Dean of the School of Advanced Translation and Translation, Beijing Language and Culture University. Photo by Li Yang
Liu Heping, Honorary Dean and Professor of the School of Translation and Translation of Beijing Language and Culture University, Doctor of Translation Studies at the Ecole Supérieure de Translation in Paris, recipient of the Chevalier de l'Ordre des l'Ordre des Edérations de l'Edre de l'Edre de la Edération de la Palm de la Edère, and an authoritative expert in the field of translation in China. She is the project leader for "Cross-language Termbase Construction and Application Development for the Winter Olympics". His research interests include translation studies, translation theory and teaching, and Sino-French cross-cultural studies.
Source: China News Service
Process Editor: TF032