Developing a Generative AI Model Using Rhetoric Prompts - Focusing on the Comparison of Rhetorical Writing and Rhetorical Prompts -

Recently, Generative AI such as ChatGPT and Midjourney have been actively used in the field of indoor architecture and are being conducted in education and practice. However, there have been many studies mainly on tools that generate content or images for data and algorithms, and the importance of prompts in the interaction between users and artificial intelligence has been relatively less discussed. Therefore, in this study, we would like to study the possibility that AI can communicate with users using prompts that mimic rhetorical writing techniques and compare the two generative methods to increase the creative potential and response accuracy of AI models. This study compared the difference between the concept, keyword, spatial element, and reference image generated by rhetorical prompts using artificial intelligence and rhetorical writing, which are used to effectively convey the designer’s intention and derive ideas in the indoor architecture design process, and analyzed the differences in terms of conceptual expression, keyword efficiency, spatial element suitability, creativity, concreteness, consistency, and predictability. Most of the time, human rhetorical writing based on one's own experience, knowledge, and emotions produced better results in creativity, but rhetorical prompts using artificial intelligence showed better results in consistency, concreteness, and predictability. These results show the possibility that collaboration using humans and artificial intelligence can achieve high efficiency in indoor architectural design. In the future, it is expected that more creative and persuasive design results can be derived by efficiently combining human sensory creativity and logical consistency of artificial intelligence.