Will AI Take over Translators’ Jobs? – A Human Translator as an Added Value in the Case of Legal Translations

With the development of large language models and artificial intelligence (AI), the past few years have seen a dramatic improvement in the quality and reliability of machine translations of legal texts, particularly in the field of neural machine translation (NMT). Translations of legal texts, particularly those that do not require authentication and thus do not need to be produced by a licensed court translator, are often done by people without a degree in translation studies. At best, they have graduated from linguistics-oriented fields, including language education; at worst, they have no formal training in translation or linguistics at all. This is often the case for law students who are required to produce translations for the law firms which employ them. If the prevailing view is that anyone can translate legal texts using AI technologies and neural machine translation, which direction will a trained human translator need to take in order to retain their position in the labour market? From my perspective, it will be the path of specialising in a particular field and ensuring firstrate results. In the case of legal translations, in addition to a proficient command of the relevant languages, it will be a competitive advantage to have profound knowledge of the legal systems of both the source and target languages, and to be able to translate texts so that they sound natural to native speakers of the target language. With a few practical examples of translations between Czech and English, which is the author’s field of expertise, this paper analyses the performance of AI-driven engines (DeepL and ChatGPT) and demonstrates that the resulting translations are not perfect and there is still room for human translators to add value, even though they need to adjust to the new reality and accept the fact that the use of NMT and/or generative AI will become indispensable, as it will allow them to satisfy clients’ demands for translations within a much shorter period of time.