AI in specialised translation: Terminology, quality and the evolving role of human translators

Abstract

Generative AI systems are reshaping the position of specialised translation by performing tasks usually associated with professional translators. This paper examines whether generative AI can handle specialised terminology at levels of expert accuracy, how its translation accuracy fares alongside that of the human professional, and how, instead of replacing, AI may transform the role of translators. Preliminary findings appear to show that while models such as GPT4 achieve a high terminological accuracy and are approaching mid-range human translation quality, they still require human supervision for nuances in domainspecific terminology. Rather than rendering the human translators obsolete, these AI tools are far more likely to transform translation work into a collaborative workspace which uses the efficiency of AI alongside human professionals. The paper ends with the need for new and targeted types of professional training and ethics to allow for this human-AI model in specialised translation.