Analisis Kesalahan Berbahasa Indonesia pada Respons Chatbot AI

Abstract

This research aims to analyze linguistic errors in AI chatbot responses, namely ChatGPT and Gemini. There are 3 aspects that are the focus of this research, namely spelling, diction, and sentence structure. This research uses a descriptive qualitative method with a content analysis approach. The research data is in the form of text generated from interactions with both types of AI chatbots, namely ChatGPT and Gemini, based on predetermined prompts. Analysis was carried out by identifying and classifying linguistic errors that comply with the Indonesian Spelling Rules (EBI). The results showed that both chatbots still produce linguistic errors, especially in the use of diction and sentence structure, while errors in spelling tend to be fewer. Meanwhile, there are differences between these two AI chatbots, especially in terms of accuracy and also consistency of language use. This research concludes that although AI chatbots help in generating sentences or text, it is still necessary to be reconsidered and refined in order to achieve optimal linguistic accuracy.