Evaluating ai-based translation in healthcare: A functionalist analysis of ChatGPT's performance

The aim of this study is to evaluate the performance of artificial intelligence-based translation tools like ChatGPT in the field of healthcare within the framework of the functionalist approach. Access to accurate and comprehensible information in healthcare is crucial for patient awareness and treatment processes. The safety of patients and their access to reliable information can be jeopardized due to mistranslations. In this thesis, ChatGPT's performance in translating informative texts in the healthcare domain is analyzed within the framework of the Multidimensional Quality Metrics (MQM), developed by TAUS and is widely used for evaluating machine translation quality. The study aims to examine the strengths and weaknesses of ChatGPT in selected healthcare texts and evaluate the impact of prompts on translation outputs. Furthermore, the need for human control, the types of errors encountered in translation outputs, and the potential risks of these errors in the healthcare field are addressed. This study seeks to contribute to the literature on the use of artificial intelligence tools for translation purposes and to shed light on the evaluation of translation quality from a functionalist perspective. Additionally, the thesis aims to highlight the advantages and disadvantages of artificial intelligence-assisted translation tools and emphasize the importance of human control for reliability and accuracy in medical translations. Keywords: AI-based Translation, ChatGPT, Medical Translation, Functionalist Approach, MQM, Error Annotation