[Moneyverse] What Do Translators Do in the Age of AI?

"Life 3.0" by Max Tegmark, published in Korea in 2017, was the first book I translated. While I diligently translated several more books, word by word, large language model (LLM) artificial intelligence (AI) suddenly appeared on the scene.

AI instantly surpassed traditional machine translation. One of the main criteria for evaluating translation quality was the "Winograd Schema Challenge." To put it simply, it measures whether the system translates pronouns accurately. In this sentence, "they" refers to the city councilmen: "The city councilmen refused the demonstrators a permit because they feared violence." When I ran this sentence through a neural machine translation engine in 2017, one of the outputs was, "The city councilmen refused the permit because the demonstrators feared violence." In contrast, by 2023, AI rendered it far more accurately: "The city councilmen rejected the demonstrators' permit application due to concerns about violence."

MTPE (Machine Translation Post Editing) refers to dividing roles, with AI handling the initial translation and humans editing the AI-translated text. As AI's translation competence is increasingly recognized, MTPE has become the dominant approach in the translation industry.

Traditional translators also did "editing." As AI translations became more refined, the importance of this role grew. When editing English texts, there are several types of edits. One is structural transformation. For example, a clause joined by a relative pronoun in English can be separated into its own sentence.

There is also the task of adapting English expressions that do not exist in Korean into more natural Korean equivalents. For instance, "Because they have a strong anti-government stance and a distorted perception of reality" can become, "Given their strong anti-government stance and distorted sense of reality." Likewise, "The spread of social media has changed the way people communicate" can be transformed into, "As social media has become widespread, the way people communicate has changed." I also correct illogical use of superlatives. For example, "He is one of the most socially influential actors in Korea," can be revised to, "He is considered a highly influential actor in Korea."

Additionally, I restructure convoluted source texts for translation. For example: "He is one of the co-founders of Palantir, which, like its namesake in 'The Lord of the Rings,' refers to an indestructible stone that can see the future," can be reworked as, "He is also a co-founder of Palantir, a company named after an object from 'The Lord of the Rings.' In the novel, the Palantir is an indestructible stone that predicts the future."

The final role is correcting errors in the original text. Readers who discover mistakes in the translated book will blame the translator. To avoid such unfair accusations, I read the original text with extra care—checking the accuracy of the content, consistency of narrative, and proper construction of sentences in context.

Since AlphaGo, AI has been transforming the way humans work, and translation sits at the heart of this change. Translators in the AI era need not only comprehension and writing skills, but also precise reasoning abilities. The work of translators who think precisely will not disappear, even in the age of AI—at least for the foreseeable future.

Baek Woojin, Economic Columnist

This content was produced with the assistance of AI translation services.