| Title |
Comparative Analysis of Prompt Engineering Reasoning Structures in Generative AI?Based Urban Design - Zero-Shot, Chain of Thought and Tree of Thought Approaches |
| Authors |
김경동(Kim, Kyungdong) ; 조태용(Cho, Taeyong) ; 정미식(Jeong, Misik) ; 조주은(Cho, jueun) ; 김세훈(Kim, Saehoon) |
| DOI |
https://doi.org/10.38195/judik.2026.06.27.3.55 |
| Keywords |
생성형 AI; 추론구조; 멀티스텝추론; 생각의나무; 도시설계 Generative AI; Prompt Engineering; Chain of Thought; Tree of Thought; Urban Design |
| Abstract |
The application of generative AI is expanding rapidly, yet its use in urban design remains limited. This study compares how three reasoning structures Zero-shot, Chain of Thought (CoT), and Tree of Thought (ToT) influence problem recognition and strategy formulation in urban design, using Seongsu Station as a case study. Zero-shot generated diverse ideas but lacked logical grounding. CoT compressed the problem into a single causal structure, converging on a focused spatial redesign strategy. ToT reframed the problem as a systemic urban overload, deriving a multi-layered strategy combining facility, operational, and structural interventions. This study demonstrates that reasoning structure influences how urban problems are defined and strategies are formulated, suggesting generative AI as a potential exploratory tool in urban design practice. |