Exploring the potential of artificial intelligence as a tool for architectural design: a perception study using Gaudí’s works

Other authors

Universitat Politècnica de Catalunya. Doctorat en Patrimoni Arquitectònic, Civil, Urbanístic i Rehabilitació de Construccions Existents

Universitat Politècnica de Catalunya. Departament de Representació Arquitectònica

Universitat Politècnica de Catalunya. QURBIS - Quality of Urban Life: Innovation, Sustainability and Social Engagement

Publication date

2023-07-22

Abstract

This study undertakes a comprehensive investigation into the comparison of designs between the acclaimed architect Antoni Gaudí and those produced by an artificial intelligence (AI) system. We evaluated the designs using five main metrics: Authenticity, Attractiveness, Creativity, Harmony, and overall Preference. The findings underline the superiority of Gaudí’s designs in terms of Authenticity and Harmony, testifying to the unique aesthetic appeal of human-created designs. On the other hand, AI-generated designs demonstrate significant potential, exhibiting competitive results in the categories of Attractiveness and Creativity. In some cases, they even surpass Gaudí’s designs in terms of overall Preference. However, it is clear that AI faces challenges in replicating the distinctive aspects of human design styles, pointing to the innate subjectivity inherent to design evaluations. These findings shed light on the role AI could play as a tool in architectural design, offering diverse design solutions and driving innovation. Despite this, the study also emphasizes the difficulties AI faces in capturing the unique facets of human design styles and the intrinsic subjectivity in design evaluations.


Postprint (published version)

Document Type

Article

Language

English

Publisher

Multidisciplinary Digital Publishing Institute (MDPI)

Related items

https://www.mdpi.com/2075-5309/13/7/1863

Recommended citation

This citation was generated automatically.

Rights

http://creativecommons.org/licenses/by-nc-nd/4.0/

Open Access

Attribution-NonCommercial-NoDerivatives 4.0 International

This item appears in the following Collection(s)

E-prints [72986]