Today, we are witnessing the advent of a new era and can foresee a radically different future due to the ubiquity of AI. Elon Musk has already predicted that, within 10 to 20 years, employment will become unnecessary and thus optional because AI will perform nearly all tasks, except, probably, for plumbing. If so, AI will rapidly replace humans in many areas.
Indeed, robots have been with us for some time. When you click on a website these days, you frequently need to verify that you are human. This is to prevent bots from accessing websites to initiate cyberattacks or for data theft. From now on, therefore, we may have to prove on many occasions that we are human. It means that it will be increasingly difficult to distinguish AI from humans in the future.
It seems undeniable that AI will take over everything soon, including the humanities and arts, and could replace teachers, researchers, scholars and artists.
AI presents as an ideal teacher: highly informative and knowledgeable, error-free and not ideologically driven. However, we may ask, “What about the human interactions between teachers and students? Can AI become esteemed mentors that students can truly revere as their milestones and guiding stars?”
AI is already showing excellent performance in visual arts, rapidly replacing visual artists. At the same time, AI is actively being used to make deepfakes, as well. If humans misuse it, AI will become harmful, not beneficial. In that sense, therefore, the humans behind AI will be more problematic than AI itself in the future.
As for AI translation, we may assume that AI could easily replace human interpreters.
Or perhaps not. A human interpreter can instantly adjust or neutralize blunt expressions that could make the other party feel unpleasant in another language. They can also omit parts that are unwittingly offensive and thus harmful for friendship or negotiation, depending on the situation. However, AI cannot be that flexible and will often interpret everything word for word, which would inevitably create problems.
When it comes to written material, AI can be a fast, good translator. Therefore, some publishers publish copyright-free classics in AI translation, instead of hiring expensive human translators. When it comes to translating documents, AI can be an excellent translator because AI translation is often crystal-clear in meaning and grammatically impeccable.
However, in the case of translating modern literature, which requires a keen literary sense and sensibility, AI may not yet be an ideal translator. The reason is that although an AI translation may be grammatically worry-free, it is often not stylish or delicate enough to carry the subtle nuance of literary works. It is especially so when translating poems saturated with profound symbolism and a heightened poetic sensibility.
Besides, AI translation is not perfect and thus not quite reliable yet. AI is prone to making mistakes because it is different from humans, who can consider many factors when translating one language into another. For example, AI often does not know when to use honorific expressions or what to choose when a word has several different meanings and usages.
Moreover, AI sometimes makes mistakes even when it translates a plain document. I tried to test it with basic information about myself. For example, the website of the State University of New York at Buffalo carries a line, “Kim has sent more than 30 of his Korean graduate students to the UB English Department for their Ph.D. degrees.” AI has translated UB into Korean as Birmingham University, not knowing that UB stands for the “University at Buffalo” in the State University of New York system. From the same website, AI translated the SNU Best Institute Director Award into Korean as “SNU Best Department Chair Award.”
In addition, AI translated “president” in “President of the LTI Korea” into 7 different Korean titles: “chongjang,” “wonjang,” “chongjae,” “daepyo,” “sajang,” “hoejang” and “isa.” Moreover, AI’s Korean translation of my bio says that I have received distinguished alumnus awards from National Taiwan University and the University of Colorado, which is not true. I have never attended those universities. Someday, AI may become perfect. Until then, we cannot wholly rely on AI translation because it can make frequent mistakes.
We often ask professors of English in Korea to choose the better result from two translations: one by a human translator and one by AI. In that case, the judges should be native speakers of English or Korean professors who can command English as well as a native speaker, because a professor of English is often an expert on English literature, not the English language.
We cannot underestimate the capability of AI, which has infinite possibilities, but AI translation is not 100 percent trustworthy yet, especially when it comes to literature. That is why we need skillful human translators until AI translation reaches perfection.
Kim Seong-kon
Kim Seong-kon is a professor emeritus of English at Seoul National University and a visiting scholar at Dartmouth College. The views expressed here are the writer‘s own. -- Ed
khnews@heraldcorp.com
