Comparison of translation quality produced by AI translator application for sarcasm in script writer presented by Val Resa on meetnitelive_metrotv

Abstract

This study aims to compare the quality of sarcasm translations produced by artificial intelligence (AI)-based translator applications to comedy scripts written by a script writer and performed by Val Resa in the Meet Nite Live program on Metro TV. The main focus lies on the ability of AI to capture the nuances of irony, subtle sarcasm, and cultural contextual meanings, which are the characteristics of sarcasm. The research method uses a descriptive qualitative approach with a comparative analysis technique on the translation results of several popular AI applications such as Google Translate, DeepL, and ChatGPT. The results of the study show that most AI applications have difficulty in translating sarcasm accurately, especially when the humour relies on local cultural context, wordplay, and emotional intonation. Some translations actually become literal and lose the desired rhetorical effect. Thus, this study highlights the limitations of AI translators in handling sarcastic humor and suggests the need for human intervention or deeper cultural context-based training to improve translation accuracy.