ChatGPT Arrives: The Language Industry Feels the First Tremor

On November 30, 2022, OpenAI released ChatGPT, a conversational AI. Within days it passed a million users — it writes emails, code and essays, and yes, it translates. Almost overnight, everyone in the translation industry was asked the same question by clients, friends and family: "Will this thing replace translators?"

First industry observations were mixed. On one hand, ChatGPT's fluency is genuinely striking: its translations often read more smoothly than traditional MT, and it can adjust tone, register and formatting on instruction, even explain its translation decisions — capabilities no traditional MT engine ever had. On the other hand, early testing quickly exposed problems: it confidently fabricates facts ("hallucinations"), is unstable on specialist terminology, supports low-resource languages unevenly, and cannot guarantee consistency across a long document.

Veteran practitioners' preliminary judgment converged: this is not "the end of MT" but "a paradigm upgrade of MT" — from dedicated translation engines to general-purpose language models. It will not immediately replace professional translators, but it will rapidly change how they work.

In the last month of 2022, the language industry stood at the doorway of a new era. What lies behind the door — 2023 will give the first answers.