Journal of the Architectural Institute of Korea (대한건축학회논문집)

Architectural Institute of Korea (대한건축학회)
권호 표지
DOI QR코드

DOI QR Code

Experiments and Evaluation of Architectural Form Generation through ChatGPT

ChatGPT를 활용한 건축 형태생성 실험 및 활용성 평가

  • Received : 2024.03.17
  • Accepted : 2024.06.04
  • Published : 2024.06.30
⟨ Previous Next ⟩

Abstract

This study focuses on utilizing ChatGPT, a generative artificial intelligence platform, for the experimentation and evaluation of architectural form generation. To achieve this objective, Grasshopper Python, serving as the linkage between ChatGPT and Rhino, is employed. The process of form creation utilizes two linguistic expressions: everyday language and mathematical language, encompassing 1) the derivation of basic modules, 2) the transformation of modules, and 3) the repetitive use and combination of these modules to generate forms. The conclusions drawn from this study are as follows: First, the study holds significant importance as it has managed to create meaningful 3D objects through the dialogue between users and objects, grounded in the core concept of hyper-connectivity, a key notion of the Fourth Industrial Revolution. Second, despite the potential for unintended results due to the limitations of linguistic expression, the study positively acknowledges the opportunity it provides for users unfamiliar with scripting to access and engage with scripts. Last, the creation of forms through mathematical expression aligns well with the attributes of scripts, and the organic use of Grasshopper has confirmed the potential for deriving forms that can be applied in actual architectural projects. These research findings suggest new possibilities for the creative use of artificial intelligence platforms in the field of architecture, offering significant insights for future architectural design and form generation.

Keywords

Acknowledgement

이 연구는 2024년도 영남대학교 학술연구 조성비에 의한 것임. 연구과제번호: 224A061006

References

  1. Bsharat, S., Myrzakhan, A., & Shen, Z. (2024, January 18). Principled instructions are all you need for questioning LLa MA-1/2, GPT-3.5/4. https://arxiv.org/abs/2312.16171
  2. Jeong, H., & Lee, J. (2023). An approach to spatial design visualization through design style combinations based on image-generation AI, Journal of the Housing Association, 34(6), 17-24.
  3. Jo, H., & Lee, J. (2023). Approaches to building facade design reflecting local identities using generative AI, Journal of the Korea Institute of the Spatial Design, 18(7), 361-370.
  4. Ko, J., Ajibenfun, J., & Yan, W. (2023, August 1). Experiments on generative AI-powered parametric modeling and BIM for architectural design. https://arxiv.org/abs/2308.00227
  5. Lee, D., & Ko, S. (2023). Experiment and evaluation of architectural image generation through artificial intelligence-based text image generation tool, KIEAE Jorunal, 23(5), 13-22.
  6. Legendre, G., & Grazini, S. (2011). Pasta by Design, London: Thames & Husson.
  7. Wikipedia. (2024, June 3). Python(program language). https://en.wikipedia.org/wiki/Python_(programming_language)
  8. Yoo, Y., & Lee, J. (2023). Generative AI-based construction of architect's style-trained models and its application for visualization of residential houses, Design Convergence Study, 22(6), 103-116.