생성형 AI 챗봇을 활용한 문학연구의 가능성과 한계 ― ChatGPT-4.o 기반 장대(張岱) 챗봇 ‘대화(岱話)’의 제작과 적용 사례

Abstract

The study explores the possibilities and limitations of utilizing generative AI chatbots, specifically ChatGPT-4.0, in literary research. By creating a persona chatbot named ‘Daihua’ based on the late Ming Dynasty literati Zhang Dai and applying it to author studies, the practical application of this research was analyzed. The chatbot ‘Daihua’ was designed to respond in a manner reflecting Zhang Dai’s literary style and tone, providing researchers with an experience akin to conversing with the actual author. The results demonstrated that generative AI chatbots could serve as valuable tools for quickly retrieving literary materials and efficiently providing the necessary information to researchers. Additionally, by recreating dialogues between the author and related figures through the multi-persona technique, the study stimulated researchers' imaginations and enabled in-depth research. However, issues such as the variability of responses and the bias in training data were identified. Moreover, there were limitations related to hallucinations and the need to rely on prompt engineering to mitigate them. In conclusion, generative AI chatbots hold significant potential as innovative tools for literary research. Future research should focus on specific solutions to these problems and further explore the potential of AI chatbots through various cases across the literary field.