Application of Artificial Intelligence-Generated Content (AIGC) in architectural design has increased rapidly. However, contemporary interpretations of traditional architectural heritage produced through such approaches often remain visually imitative, resulting in compromised cultural integrity and reduced semantic accuracy. To address this issue, this study proposes a traceable and constraint-driven framework that transforms architectural heritage knowledge into reproducible AI-assisted design outcomes, using Huizhou-style architecture in China as a case study. Methodologically, the cultural genes of Huizhou architecture are extracted and encoded through grounded theory. The priorities of these genes are then quantified using the Analytic Hierarchy Process (AHP) and subsequently translated into weighted constraints embedded within AIGC prompts. These constraints are further structured through a three-layer symbolic transformation framework encompassing morphology, semantics, and behavior. The generated outcomes are empirically evaluated using the User Experience Questionnaire (UEQ) and paired-sample t-tests. The results indicate that the proposed workflow significantly enhances users’ perceived cultural accuracy, emotional resonance, and overall perceived quality of AI-generated outputs. This research presents a reproducible AI collaborative workflow that contributes to sustainable and verifiable heritage translation within intelligent design environments.
Feng et al. (Fri,) studied this question.