How Well Can AI Contribute to Interior Architecture? A Comparative Analysis of Descriptive Accuracy

Read full text Summarise with VIRAT
Authors: Yaren Şekerci, Müge Develier
Cited by: 0
References: 35

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This study evaluates the performance of three AI models—Claude 3.5 Sonnet, Gemini 1.5 Flash, and ChatGPT 4o—in generating interior design outputs based on seven key design criteria: design style, color, lighting, furniture and product, interior materials, architectural features, and spatial layout. The evaluation was conducted across six different space designs, with 15 participants scoring the AI-generated outputs on a scale of 1 to 5. The results indicate that Claude 3.5 Sonnet achieved the highest overall performance due to its consistent scores across multiple criteria, followed closely by Gemini 1.5 Flash, which excelled in design style and color but exhibited slight variability. ChatGPT 4o, while demonstrating strong performance in furniture and lighting, struggled with inconsistencies, leading to its lower overall ranking. Despite their competitive performance, areas such as spatial layout and interior materials presented challenges for all models, highlighting opportunities for improvement. This study underscores the growing potential of AI in supporting design processes while emphasizing the need for further refinement and expansion to address limitations in complex spatial and material contexts.

Fields of Study & Subjects

Architecture and Computational Design Architecture Design and Social History Artificial Intelligence Applications

Publication details

Journal / source
Journal of Computational Design
Volume
7
Issue
1
Pages
95–132
Publication date
30 March 2026
Language
TR