APSIRI Research: From “Correct” to “Great” — Translators in the Age of AI

Machine translation has crossed a threshold. Systems such as ChatGPT and DeepSeek now produce translations that are fluent, fast, and often impressively accurate — raising urgent questions about the future of the translator’s craft. A new research outcome from the Asia-Pacific Social Innovation Research Institute (APSIRI), led by Ling Zou with Jingjing Zhao and Ruolan Lai, addresses these questions directly, moving the conversation “from correct translation to great translation” and examining the digital intelligence literacy translators now require.

The study’s framing is itself a contribution. As machine systems become highly capable of producing “correct” translations — renderings that are accurate at the level of meaning — the distinctive value of the human translator shifts. Correctness becomes a baseline that machines can often meet; what remains harder to automate is greatness: the sensitivity to nuance, tone, register, cultural resonance, and rhetorical effect that distinguishes a merely accurate translation from a truly excellent one. The research asks how human and machine can best work together to reach for that higher standard.

To ground its analysis, the study undertakes a comparative examination of translations produced by ChatGPT, DeepSeek, and human translators. This comparison reveals both the considerable strengths of contemporary AI systems and the areas where human judgement continues to add distinctive value. The picture that emerges is not one of replacement but of complementarity: machines and humans bring different capabilities, and the most powerful results may come from their thoughtful combination.

Central to the study is the concept of “digital intelligence literacy” — the set of capabilities translators now need to work effectively alongside AI tools. This is more than knowing how to operate a piece of software. It involves understanding what these systems do well and where they fall short, knowing how to prompt, evaluate, and refine their output, and retaining the human judgement required to take a translation from correct to great. Far from making the translator obsolete, the rise of capable AI redefines and, in important ways, elevates the translator’s role.

The implications reach well beyond translation. Across many fields, capable AI tools are becoming collaborators rather than mere instruments, and the question of how to work well alongside them — what literacy, judgement, and skill that requires — is becoming central to professional life. The study offers a thoughtful model for thinking about human–machine collaboration in any domain where AI can produce competent work but human excellence still matters.

For the translation profession specifically, the message is encouraging but demanding. The future belongs not to those who compete with machines at tasks machines do well, but to those who develop the literacy and judgement to direct, refine, and surpass machine output — to reach for greatness that correctness alone cannot achieve.

This research advances APSIRI’s work on inclusive technology and knowledge dissemination. By examining how human capability and artificial intelligence can best combine, the Institute seeks to support a future in which technology amplifies human skill and judgement rather than displacing it.