設計基因之AI生成圖像在皮爾斯符號學的應用

Application of Design DNA's AI-generated images in Peirce's Semiotics

本研究探討人工智慧（AI）生成圖像在皮爾斯（Peirce）符號學理論中的應用，並結合設計基因概念，建構AI生成圖像的符號學理論架構與設計應用模式。本研究以Midjourney生成像工具為例，分析AI生成圖像的符號學運作機制，提出其創作過程可視為皮爾斯符號三元關係（客體—詮釋—表徵）的動態送代歷程。此外，本研究引入設計基因概念，其中主設計基因作為設計的固定條件，從屬設計基因作為設計的可變條件，以此作為AI生成圖像的指令編碼基礎。為驗證其可行性，本研究以「只有蕨」品牌產品為案例，作為設計基因拆解與圖像重構的設計實踐樣本，應用AI生成圖像工具重新生成椅子造型設計。結果顯示：主設計基因確保視覺一致性，從屬設計基因則擴展設計多樣性；圖像與文本雙重輸入可提升生成可控性，而設計基因編碼程式（如A+a\b\c\d）能系統化引導風格化輸出，驗證皮爾斯「無限符號」理論於AI設計中的連續詮釋性。本研究以皮爾斯符號學、設計基因與AI生成圖像技術為基礎，提出一套結構化的AI設計指令方法，為設計領域提供理論依據，並展現AI在人機協作、品牌識別、產品設計與設計方法論發展等領域的應用潛力。

This study investigates the application of artificial intelligence generated imagery within the framework of Charles S. Peirce's semiotic theory, integrating the concept of ''Design DNA'' to construct a semiotic model and design methodology for AI-generated images. Using the Midjourney image generation tool as a case study, the research analyzes the semiotic mechanism behind AI image generation and interprets the creative process as a dynamic iteration of Peirce's triadic relation—object, interpretant, and representamen. Furthermore, the study introduces the concept of design DNA, where primary design DNA represent fixed design attributes, and subordinate design genes serve as variable conditions. These are used as the basis for AI prompt encoding. To validate this framework, the study adopts the product line of the brand ''Only Fem'' as a design case, deconstructing its design DNA and applying them in an image reconstruction process using AI to regenerate chair forms. The results indicate that primary design genes ensure visual consistency, while subordinate genes enhance design diversity. The use of both image and text inputs significantly improves generative control, and a design DNA encoding system (e.g., A+a\b\c\d) effectively guides stylized outputs. This supports Peirce's notion of ''unlimited semiosis'' by demonstrating iterative interpretability in AI-based design processes. Grounded in Peircean semiotics, design DNA theory, and AI generation technology, this study proposes a structured AI design instruction methodology that offers a theoretical foundation for the design field. It further highlights Al's potential in areas such as human-machine collaboration, brand identity development, product design, and the evolution of design methodologies.