Experts say large language models make advanced English more necessary
The rapid advancement of artificial intelligence (AI) is fueling renewed debate about the role of English in global communication. As AI-powered translation becomes faster and more accurate, some argue that learning English is becoming less essential.
Experts, however, warn that this perspective oversimplifies the issue, particularly for those looking to fully seize the opportunities of the digital revolution. While AI is undeniably transforming the communication landscape, they emphasize that the enduring value of mastering English remains as crucial as ever.
The root of this continuing English advantage lies in the data that powers the systems themselves.
An estimated 90 percent or more of the training data used for current generative AI systems comes from English content. This means that, while AI systems are designed to handle many languages, they often perform best and respond most naturally in English. Many technical tools and the latest research are released first in English, and the most advanced AI features are typically optimized for English communication, according to experts.
“But more importantly, AI systems are becoming accustomed to English-structured thinking. In other words, AI tends to process and generate language following conventional English grammar, sentence structure and logic, which would put those with strong English skills at a distinct advantage,” Kim Kyeong-seo, CEO at VAIV Company, a Seoul-based firm specializing in AI, told The Korea Times.
This bias is not merely a technical limitation, but a deep-seated linguistic one that affects how AI processes all other languages.
Studies show that AI models often carry English-centric biases, leading to unnatural outputs in other languages. Behind the scenes, these AI models “think” in English before rendering a translation in a target language. This can impact the quality and nuance of AI’s responses in non-English interactions, said Kim Kyung-joong, a scholar at the Gwangju Institute of Science and Technology.
“AI has removed basic language barriers. But it still doesn’t feel like we’ve reached the point where conversations with AI are as natural as those between humans,” he said. “That’s why I still encourage my students to learn English.”
Yet, other educators argue that this linguistic asymmetry does not lessen the need for English but instead fundamentally changes it.
Moon Hyung-nam, professor at Sookmyung Women’s University and chairman of the Korea Association of AI Education, says the need for English is evolving — from daily conversations to deeper, more technical collaboration with AI.
“Mastering English enables users to give more accurate instructions to AI, properly assess the work AI provides and correct any mistakes. With more possibilities opening up thanks to AI, the ability to read and understand English has become even more important,” he said.
Moon points out that the AI era divides users into those who simply depend on AI and those who actively take advantage of the technology.
“Those who can review and refine AI-generated translations, making them more natural, are able to achieve much higher-quality results than those who just use AI output as it is,” Moon said.
With that change, he said English is no longer just a functional tool — it’s now an essential language for effective collaboration with AI.
“Thanks to AI, the days when not knowing English meant a competitive disadvantage are ending. But we are entering an era where only those who understand English well can fully harness the power of AI,” Moon said. “For this reason, the goal of English learning will shift from communication to developing global thinking skills through AI.”
