Will AI Replace Human Translators?

The Early History and Evolution of Machine Translation

The history of machine translation technology dates back to the 1940s Cold War era, when the United States and the Soviet Union began developing machine translation technologies to decipher each other's confidential documents. In 1951, Warren Weaver, a researcher in machine translation, was appointed as the chairman of the National Science Foundation (NSF), leading to vigorous development of machine translation technology at major universities like the University of Washington, the University of Michigan, and UC Berkeley. However, at that time, machine translation technology required manual input of all rules, making it challenging to translate various languages.

Innovation in Artificial Intelligence Based Translation

The advent of ‘Artificial Intelligence (AI)’ marked a new chapter in the use of machines for language translation. AI was able to quickly learn from the vast amount of language big data, which was impossible for traditional computers or translators. This led to the development of AI-based 'neural machine translation'. Neural machine translation used the principles of machine learning and deep learning to innovatively analyze and understand large amounts of language data. This technology allowed computers to capture nuances of context and produce more natural translations, not just mechanically applying language rules. In this process, AI learned from thousands of different language examples, understanding the structure, grammar, and context of languages, and providing more accurate and natural translation results. Notably, this technology didn't rely on rules defined by humans but learned patterns from massive language data, effectively handling the complexity and diversity of languages. This significantly improved translation accuracy and opened up the possibility of translating various language combinations. Consequently, AI-based neural machine translation marked a revolutionary turning point in the field of language translation, leading to technological advancements that enabled faster and more accurate global communication.

The Contemporary Application and Impact of AI Translation

The remarkable development of AI translation has catalyzed the use of AI in place of human translators across various fields. This trend is particularly noticeable in businesses, government agencies, and educational institutions. For instance, global companies are using AI translation services to translate their websites and product documents into multiple languages, reducing costs and increasing work speed. Government agencies are actively utilizing AI translation tools for official documents and international communication, significantly improving efficiency. In the field of education, online education platforms are using AI to translate course content into various languages, providing broader accessibility to learners worldwide. Similarly, language learning applications are offering real-time translation and language learning tools through AI, bringing changes to traditional language learning methods. In the travel and tourism industry, AI translation plays a crucial role. Mobile apps with real-time translation features enable travelers to easily understand foreign language signs, menus, and information boards, making travel enjoyable without language barriers. Additionally, in customer service, AI chatbots and automatic translation systems can swiftly respond to customer inquiries in multiple languages.

Will AI Replace Human Translators?

The question of whether AI can replace human translators has been heating up in our society since the introduction of ChatGPT. The translation capability of AI has grown remarkably. ChatGPT performs language translations more naturally than previous tools and can communicate with people. Despite improved performance, there are still controversies and issues regarding the level of translation. So, why can't AI completely replace human translators?

The constant evolution and change of language are major reasons why AI cannot fully replace human translators. Language is not just a collection of codes but a vibrant system that continually changes with time and culture. New vocabulary and expressions are constantly emerging, and existing words and grammar structures also change over time. For example, technological terms or contemporary cultural ideas that have emerged in the last 10-20 years did not exist before. Humans are in a much better position to respond to these language dynamics than AI. Human translators can quickly recognize new linguistic phenomena and cultural contexts and reflect them in translations. On the other hand, AI translation systems depend on massive data and algorithms, limiting their ability to immediately grasp and apply these rapid language changes. AI learns from past data, so understanding and capturing the latest language phenomena, slang, or neologisms can take time.

The importance of context in the translation process is another key reason why AI cannot completely replace human translators. Despite significant advancements in AI translation technology, human translators are still essential for complex texts or content requiring cultural nuances. According to Ethnologue, there are over 7,000 languages worldwide, each with unique characteristics and contexts. This linguistic diversity includes expressions and meanings that can only be understood correctly through context. Cultural context or specific social situations require a human's sensitive understanding to accurately capture and translate the intended meaning. AI still struggles with understanding and processing linguistic subtleties, including metaphor, simile, exaggeration, hidden meanings, neologisms, and ambiguity. These sentences need to be interpreted differently based on the speaker's intent and context, which is challenging for AI to ascertain appropriately. Even with advanced technologies like deep learning, replicating this linguistic complexity perfectly remains a challenge for current AI. AI translation systems tend to treat sentences or words as independent elements, often overlooking or misinterpreting complex contextual elements such as relationships between sentences, cultural background, or the author's intent. For instance, in business and particularly in the medical field where extreme accuracy and reliability are required, incorrect translations can lead to serious consequences. In medical situations, translation errors can directly impact patient safety and, in some cases, be life-threatening. Accurate translation of medical terminology, clear communication of medical guidelines, and detailed explanations of a patient's medical history and symptoms are crucial. Medical terminology is highly specific and can be interpreted differently based on context, making it a challenging task for AI. Additionally, in the medical field, the same term can have different meanings depending on cultural background and individual health status, and it is important to accurately understand and translate these nuances. AI has limitations in recognizing these subtle differences, leading to serious errors due to incorrect information transmission. Therefore, in business sectors where accuracy and reliability are crucial, such as the medical field, the role of human translators remains important. Human translators can accurately understand and translate complex medical information through their expertise and experience, thereby minimizing the negative outcomes of incorrect translations. This highlights the need for human translators in areas where AI translation has not yet achieved the required accuracy and sensitivity.

HUMAN IN THE LOOP

Even though AI translation cannot yet completely replace human translators, there are many reasons to use AI translation. Humanity has no reason to reject the efficiency of AI, which will inevitably be deeply integrated into civilization. This underscores the necessity of human interaction in the modern field of translation. The concept of "HUMAN IN THE LOOP" emphasizes the interaction and cooperation between AI and human translators in contemporary translation. It is based on the idea that AI can achieve higher-quality results under the continuous intervention and supervision of human experts, rather than working independently. AI translation excels at processing large amounts of data and performing general translation tasks, but human translators provide the knowledge and experience needed to understand and handle complex linguistic issues and nuances that AI has not yet solved. In the HUMAN IN THE LOOP approach, translators are responsible for reviewing, modifying, and improving translations generated by AI. Through this process, translation quality can be significantly improved, and AI can learn and develop based on human feedback. This approach does not simply change the role of human translators from translators to reviewers and editors but emphasizes their important role as collaborators and interactive members in the translation process with AI. Ultimately, the HUMAN IN THE LOOP approach presents a future model of translation where AI and humans complement each other's strengths and cooperate. This opens the way to overcome the limitations of AI translation technology and utilize the full capacity of human translators to provide a higher level of translation service.

The Future and Challenges of AI Translation

The case of a Japanese individual translating a Korean webtoon into Japanese and winning an award in February 2023 highlighted the role of translation AI. This ordinary 40-year-old housewife, Yukiko Matsushita, won the award despite her limited Korean language skills, bringing various issues triggered by the development of AI to the surface. The Korean Literature Translation Institute added a requirement to the 2023 Korean Literature Translation Newcomer Award competition guidelines stating, "Joint translation with a person or machine is not allowed, and if plagiarism of another's work is confirmed, the award will be revoked." However, there are opinions questioning the effectiveness of these new guidelines. "It's understandable to exclude AI translation to assess the qualifications of a translator, but if a candidate uses an AI translator and then refines it several times so that it's not noticeable, how realistically can this be discerned?" This controversy presents a special case that shows many aspects at once, from redefining translation in the AI era to the scope of collaboration with AI, regulations on joint creative works, copyright, and the need for legal and social system adjustments and ethical considerations. This surprising but predictable anomaly signifies that coexistence with AI is an unstoppable reality and the future. The remaining question is how well we can use this 'tool'.