How will AI impact the Future of Translation Industry

An exclusive Interview with Liu Heping, former dean of the Graduate Institute of Interpretation and Translation of Beijing Language and Culture University

In recent times, there has been a continuous stream of advancements in AI technologies. AI-powered translation applications and software are gaining widespread usage. In addition to well-known translation apps like Google Translate, newer AI tools like ChatGPT are being employed for language translation, delivering even more precise and fluent translations.

What will be the impact of AI on the translation industry? The East Meets West program conducted an exclusive interview with Liu Heping in Paris to discuss this matter. Liu Heping, the esteemed honorary dean of the Graduate Institute of Interpretation and Translation at Beijing Language and Culture University, shared her insights on this topic:

China News Service Reporter: We have seen more and more AI applications in translation, especially the growing influence of ChatGPT and other new AI technologies. Will AI significantly impact human translation in the future?

Liu Heping: In my view, the development of AI can be compared to the progress of the mobile phone. When mobile phones first came out, they were only capable of making calls. As technology advances, there are more and more functions. However, mobile phones still do not replace humans despite its advancement. The same goes for AI. ChatGPT may have an impact on human translation, but in fact, such technologies offer more help to humans rather than replacing them. Humans also adapt to the trend.

AI can provide more translation services to humans. As people have different levels of proficiency in translation, AI will replace translators whose translation is not better than AI. We need more advanced translators in the future, which means that the human translation should not only capture basic information, but also convey its emotion.

I have already used ChatGPT for translations many times. It can translate texts smoothly without grammatical errors. However, AI generated translations are banal. I’d like to express different emotions from the source text, which cannot be done by AI at present. Human translation should leave some room for ‘creativity’ and ‘emotion’.

China News Service Reporter: However, there are still lingering concerns among the public regarding the potential complete replacement of human translators by AI.

Liu Heping: MTI (Master of Translation and Interpreting) Students, parents as well as the public are all concerned. I think it is unnecessary. AI translations will never replace human translators. In the early years, we used paper sheet to write articles, then we used computers and voice recognition technologies. Our productivity increases with technological advancement. People are definitely unwilling to return to the old days when we wrote articles on paper sheets. So, I suggest students to embrace future technologies and unleash human values.

Translation softwares might carry out 90% of future translation. However, we still need more advanced human translators to handle the remaining 10%. We always need human translators to complete the final part of the translation, imposing higher standard on professional development. Since AI technologies are constantly changing, I am still studying and discovering translation even though I am retired.

I am reflecting on the limitation of ChatGPT and evaluating my own strengths after using it in person. As a result, I came up with the conclusion that AI translation can’t replace human translators.

Photo credit: Vision China

China News Service Reporter: Will AI replace your job in the field of literary translation?

Liu Heping: While AI can translate some literary texts, the translation is banal since it failed to capture emotions conveyed by the author. Thus, I advised my students not to be afraid of technological progress. Instead, they should focus on enhancing their capabilities and learning how to effectively harness the power of AI. This approach would promote both quality and efficiency in translation processes.

China News Service Reporter: China International Communications Group cooperated with iFLYTEK CO.LTD. to build an AI translation platform in 2018. The United States also invested greatly on such platforms. What do you think of these AI translation platforms based on your understanding?

Liu Heping: I used to utilize Google Translate frequently to translate among Chinese, English, and French. Some foreign translation software are good at translating among different European languages. Today, students use DeepL more with good quality. However, more discussions are needed on how good the DeepL translation is. Generally, AI translation platform performs well in translations among different European languages. However, it’s not doing well when translating between Western and Eastern languages. Primary issues include redundancy, coherence, and the logical consistency among sentences. Although it does not have major grammatical errors; the expressions are still unnatural.

We conducted some translations with Microsoft and discovered that AI translation platforms can learn themselves. Prior to the conference, we put the speaker’s presentation to the platform so that it gains basic understanding on the text, resulting in a significant reduction of AI translation error rates during the conference. This is comparable to human translation. But AI translation platforms are still unable to translate accurately and provide multilingual simultaneous interpreting service.

During the 2019 Australian-China (Melbourne) Economic & Trade Expo, held in July 2019, iFLYTEK CO.LTD showcased its machine translation products, smart office laptops, and intelligent robots from its partners. This demonstration aimed to highlight the advancements and capabilities of iFLYTEK and its partners in the field of AI. Photo credit: Zhang Xiaoli.

China News Service: ‘The establishment and application of the cross-language term ase for the Winter Olympics’ centered around the 2022 Beijing Winter Olympics and Paralympics, including term ase in eight languages: Chinese, English, French, Japanese, Korean, Russian, Arabic, and Spanish. As the project leader, why do you establish this termbase? Does it meet your expectation? Will it apply in future events?

Liu Heping: Beijing Organizing Committee for the 2022 Olympic and Paralympic Winter Games launched the linguistic action plan for Beijing Winter Olympics. Beijing Language and Culture University initiated relevant projects to serve in the Winter Olympics. In 2021, we delivered the third version of ‘Cross-Language Termbase for the Winter Olympics’, which played an important role in 2022 Beijing Winter Olympic and Winter Paralympic events.

The research and development of the cross-language termbase is a ground-breaking project. Before the 2022 Beijing Winter Olympic, technical materials related to language used in previous Olympic Games were hard to find. The Beijing Winter Olympics presented an excellent opportunity to develop such cross-language termbase. We established expert panels for 8 languages as confirmed by IOC for the Beijing Winter Olympics, with regular meetings to advance termbase development and complied a total of 132,000 terms across the eight languages in the end. This is the first multi-language smart platform around the world that covers core terms, operation terms, and common glossaries, thus filing the gap in Winter Olympic terms.

We hope the termbase to be the linguistic legacy for international Olympic games following the Beijing Winter Olympics, with potential for application and adaptation in future events. I personally hope that this term base expands its scope and serves as a national platform for specialized cross-language sports terminology, with the support of the Chinese government in due course. For example, this platform could encompass sports such as football, tennis, marathon races and so on. If we intend to utilize the term base to other sports events like the Hangzhou Asian Games and the Paris Olympics, relevant content must be added. The approach for building the platform is already clear, and there is potential for further dissemination in the future.

China News Service Reporter: Can this terminology database be utilized in other AI contexts?

Liu Heping: Yes! Cross-Language termbase for the Winter Olympics is more than a glossary platform. It will present like an encyclopedia and offer relevant contexts when searching for a word, so that we can understand the context of this word. In addition, Beijing Language and Culture University and Tsinghua University are also conducting research projects on AI. We hope to integrate this research with the termbase to maximize its effectiveness. Of course, this will require continuous investment in the future.

Guest Profile

Liu Heping is the honorary dean of the Graduate Institute of Interpretation and Translation of Beijing Language and Culture University. Photo credit: Liyang.

Professor Liu Heping, the esteemed honorary dean of the Graduate Institute of Interpretation and Translation at Beijing Language and Culture University, holds a PhD in translation studies from Ecole supérieure d'Interprètes et de Traducteurs. She has been recognized for her remarkable contributions to education with the prestigious Palm Chivalric Order in Education awarded by the Ministry of National Education of France. Renowned as a leading expert in translation studies within China, Professor Liu serves as the project leader for the Cross-Language Term bases for the Winter Olympics. Her research primarily focuses on translation studies, encompassing both theoretical and practical aspects of translation and interpreting, as well as conducting cross-cultural studies between Chinese and French.