The position of ChatGPT3 chatbot in the process of translation of written medical texts and evulation of translation product quality

Some versions of AI-based chatbots are now available to the public as open source, including certain iterations of conversational robots. However, the ChatGPT chatbot, presented by OpenAI as a substantial language model, significantly distinguishes itself from other conversational robots. From the perspective of translation studies, chatbots like ChatGPT (computer programs used for interaction) are utilized by translators both in academic settings and in their professional lives for the production of translated texts. Despite the high-quality translations produced by conversational robots like ChatGPT due to advancements in machine learning, the machine translation process is not entirely automatic. Consequently, the intervention of a human translator is required in pre-translation, during the process, and post-translation stages. More importantly, the accurate and precise translation of written medical texts is crucial for the health and treatment of patients. Therefore, the quality and accuracy of the translation product are of utmost importance. In this thesis, the position of the chatbot named ChatGPT3 in the translation process, its translation performance in three different written medical texts, and the necessity of human intervention before and after the translation process are examined. The examination results reveal that ChatGPT3 poses significant challenges in translating lengthy and information-rich texts. Additionally, raw machine translation products and machine translation products obtained through post-formatting and pre-formatting are analyzed using the BLEU metric, with human-translated products as references. Although the BLEU metric does not provide a comprehensive analysis in terms of linguistic, contextual, and accuracy aspects, the analysis results indicate that the translation products obtained from ChatGPT3 received low scores. Within the scope of all data, it is concluded that ChatGPT3 is inadequate in translating medical texts.