Artificial Intelligence for Aesthetic Assessments in Architecture

Comparing the Differences, Strengths, and Limitations of AI and Humans
1. Auflage 2027
ISBN: 978-1-041-02200-8
Verlag: CRC Press

Buch, Englisch, 288 Seiten, Format (B × H): 156 mm x 234 mm

This cutting-edge guide describes to readers the growing applications of machine learning (ML) to evaluate aesthetic design in architecture and explains how this can be implemented and coordinated with human judgment to improve building design and fabrication techniques.

Through collaborative studies on ML, form-generating methods and robotic fabrications, the authors find and develop 3D sampling models, fabricate them at full scale, and assess their aesthetic values using human surveys. Finally, they conduct a series of comparative analyses on data collected from both machine and human evaluations to identify the differences and similarities in their design assessments. The inclusion of ML for this purpose requires extensive collaboration and procedural effort involving interconnected workflows for a rigorous comparison of aesthetic judgments between ML and humans. Therefore, the authors describe these workflows, including data collection, ML, form generation, robotic fabrication, and human assessment. This approach gives readers an understanding of the compatibility between human judgment and ML models in the aesthetic evaluation of the same design objects and to compare the differences in aesthetic judgments between ML and humans. This will allow them to generate designs and make aesthetic judgments, ultimately, refining designs to enhance both their functionality and visual appeal.

This practical guide will bring practicing architects and students up to speed with the latest applications of ML in aesthetic assessment and will provide them with a strong basis to adapt to any future developments. Construction and material firms interested in supplying new types of building materials will also benefit from the description of form-generating methods and fabrication techniques.

Zielgruppe

Postgraduate, Professional Reference, and Undergraduate Advanced

Autoren/Hrsg.

Park, Jin-Ho

Kim, Jiwon

Fachgebiete

• Mathematik | Informatik EDV | Informatik Daten / Datenbanken
• Geisteswissenschaften Architektur Architektur: Berufspraxis
• Geisteswissenschaften Architektur Gestaltung, Darstellung, Bautechnik
• Mathematik | Informatik EDV | Informatik Informatik Künstliche Intelligenz Maschinelles Lernen
• Geisteswissenschaften Design Produktdesign, Industriedesign

Inhaltsverzeichnis

Introduction. Part 1: Fundamentals: machine learning models, aesthetic categories, datasets, and model training 1. Machine learning models and generative AI 2. Aesthetic categories, measures and evaluation models 3. Dataset preparation and model training Part 2: Robotic fabrications and tools: basic setting, calibrating, and fine-tuning 4. Robotic fabrications with hot wire 5. Robotic fabrications with a milling tool Part 3: Five experiments integrating form-generating method, fabrication and human participant survey 6. Parametric wall designs for a pilot study 7. Experiment with walls featuring daylight 8. Experiments with images generated from Midjourney 9. Experiment with generative AI images referencing modern architects’ works 10. Fine-tuning Stable Diffusion with LoRA 11. Quantitative comparison of human and CNN judgements. Conclusion. Appendices

Über Autor(innen)

Jin-Ho Park is Professor of Architecture at Inha University, Korea. Previously, he was a tenured Associate Professor at the University of Hawaii at Manoa, USA. His recent books include Rudolph M. Schindler’s Unbuilt Architecture (2025), Pavilions for Giving (2025), and Architectural and Urban Subsymmetries (2022).

Jiwon Kim is a postdoctoral researcher and visiting professor at Inha University, Korea. She earned her Ph.D. from the University of Tokyo, Japan and conducts research on architectural design and theory, urban revitalization, and AI applications in architecture.