인공지능 생성 콘텐츠(Artificial Intelligence Generated Content)의 가추법적 작동과 예측 프라이버시의 윤리: 챗지피티(ChatGPT) 지브리 스타일 이미지 사례 연구

Artificial intelligence generated content (AIGC) refers to the automated creation of various forms of content in response to user prompts. Beyond simply generating outcomes, AIGC operates as a mechanism through which AI algorithms structure user identity in specific ways. This study examines the generative logic of AIGC through the lens of Peirce’s concept of abduction—a reasoning process that derives the most plausible case from a given result and rule. Focusing on the case of Ghibli-style image generation using ChatGPT, the analysis shows that the prompt functions as a ‘result,’ which is combined with pre-trained ‘rules’ to produce a culturally resonant ‘case’ in image form. While these Ghibli-style outputs may appear to represent individual identity, they are transformed into repeatable emotional codes and visual conventions, recursively reused as training data that influence other users’ experiences. AIGC’s abductive operation thus reveals a structure in which user data unintentionally contributes to the prediction and shaping of others’ identities, thereby raising concerns about predictive privacy violations. This study argues that AIGC functions as an algorithmic semiotic apparatus that constructs user identity and produces cultural symbols at a collective level, and calls for a critical ethical reconfiguration of the relationship among technology, identity, and privacy.