Lee Hae-rin is a City Desk reporter at The Korea Times, covering social issues, tourism and taekwondo. She is passionate about speaking up for the rights of minorities, including women, LGBTQ+, people with disabilities and animals as well as discovering the latest makgeolli trend in town. Feel free to reach her at lhr@koreatimes.co.kr.
AI can translate words, but doesn't bear responsibility, experts say
Novelist-translator Moon Ji-hyuk, speaks during a panel at the Literature Translation Institute (LTI) Korea's International Translators' Summit in Seoul, Friday. From left are literary critic and session moderator Heo Hee, Moon, Lee Sang-bin, a professor of English interpretation and translation at Hankuk University of Foreign Studies, and translator-poet Stine An. Captured from LTI Korea's YouTube account
By Lee Hae-rin
International Translators' Summit participants stress preserving human ethical judgment while giving translators control over AI tool key
Translators must defend their role not by competing with artificial intelligence (AI) on speed, but by taking responsibility for the cultural and ethical choices embedded in their work, experts said Friday at a conference on literary translation organized by the Literature Translation Institute (LTI) Korea.
The International Translators' Summit brought together Lee Sang-bin, a professor of English interpretation and translation at Hankuk University of Foreign Studies; Korean American literary translator and poet Stine An; and novelist-translator Moon Ji-hyuk.
The discussion wasn't about whether AI can produce fluent translation. Instead, the panelists focused on who should bear responsibility for a translation, how translators should use AI and what kind of work must remain human-led.
Lee argued that translation is not simply the transfer of meaning from one language to another.
“I want to say that translation is not the work of copying meaning, but an act of responding to the words of the other,” the professor said, adding that a translator can never completely know an author or source text. However, he said, the translator must still answer the call through choice of words, tone, rhythm and cultural framing, which distinguishes human translation from AI-generated output.
“AI does not feel uncomfortable after reading a source text. It does not feel guilty about choosing a particular word, nor can it step forward and take responsibility for the result,” he said.
Lee said human translators’ apparent inefficiency should be understood as a necessary ethical interval rather than a weakness. “This time of hesitation is the time in which responsibility takes concrete form,” he said.
He proposed a future in which technological translation, oriented toward efficiency, automation and scale, coexists with ethical translation, which centers on response, responsibility and a relationship with the author or text.
A poster for Literature Translation Institute (LTI) Korea's 2026 International Translators' Summit held in Seoul, Friday / Courtesy of LTI Korea
An offered a more pragmatic account of how AI can support translators without taking over their creative work. She stressed that literary translators do much more than produce a finished text: They organize readings, communicate with publishers and institutions, apply for grants, promote writers and build communities across languages.
She described AI as a useful tool for such work — including drafting emails, preparing event materials and summarizing recordings — particularly as a translator manages multiple projects and professional commitments. AI, in that sense, can function as an assistant that frees up time for the more demanding work of translating literature and creating human connections, she said.
However, An cautioned against machine-translation post-editing models that make translators responsible for correcting text they did not create. Instead, she called for a human-centered approach in which translators retain control over the goals, tools and final decisions of a project.
Moon framed the speakers’ ideas as two paths toward the same decision: AI is disruptive not only because it is faster and cheaper, but because it exposes how narrowly translation has been defined.
“Human translation’s value is not in perfection, but in being able to explain why one translated something in a way and bear the consequences of that choice,” Moon said. “A good translation is not about producing the correct answer. It is closer to choosing one responsible failure among countless impossible choices.”
He said translators should retain three key roles in an AI-assisted future: final authority over major interpretive and ethical decisions, responsibility for relationship among authors, readers, publishers and communities, and awareness of translation’s inevitable limits.
The summit runs through Saturday and was organized to mark LTI Korea’s 30th anniversary.