Artificial intelligence (AI) technology, namely large language models (LLMs), has developed rapidly. Chat Generative Pre-trained Transformer (ChatGPT ), is one example of a technology product designed as a natural language processing model identified as GPT-4. This study aims to compare the ability of Large Language Models (LLMs), namely ChatGPT in producing natural and meaningful dialogue in podcasts with human language. To test the hypothesis that AI is capable of mimicking conversations, researchers compared podcasts featuring dialogues between humans and chatbots generated by ChatGPT. The research method used is qualitative research method to describe the difference between AI and human dialogue. The object of research is podcast data that has been published digitally, namely podcast recordings from the YouTube platform in the Turn Level Gaming (TLG) account channel and podcast transcripts generated from videos. Data collection techniques were carried out by means of documentation, namely collecting podcast video recordings and transcripts from sources available online. Data analysis techniques use content analysis to uncover fundamental similarities and differences between the two types of podcasts. The results of the analysis were compared with related literature to provide justification for the research findings. The results of content analysis show that there are elements of theme, category of content, purpose of message, and style of language. The structure of human sentences in podcasts shows social flexibility, emotional function, and the formation of shared meanings. Meanwhile, the results of narrative analysis show 3 elements, namely orientation, complications, and evaluation. The sentence structure on the cchatbotis still limited to the accuracy and efficiency of information, it has not been able to compile a contextual narrative or argumentative structure. This difference shows that human interactions with humans and chatbots differ fundamentally, not just in the content of the conversation, but in the way sentence structure is used to create meaning, relationships, and emotions.
Ningsih et al. (Thu,) studied this question.