• icon+90(535) 849 84 68
  • iconnwsa.akademi@hotmail.com
  • icon Fırat Akademi Samsun-Türkiye

> Detaylı Arama

Article Details

Article Details

  • Article Code : FIRAT-AKADEMI-14628-5821
  • Article Type : Araştırma Makalesi
  • Publication Number : 3C0196
  • Page Number : 17-32
  • Doi : 10.12739/NWSA.2026.21.2.3C0196
  • Abstract Reading : 92
  • Download : 24
  • Atıf Sayısı : 0
  • Share :

  • PDF Download

Issue Details

  • Year : 2026
  • Volume : 21
  • Issue : 2
  • Number of Articles Published : 1
  • Published Date : 1.04.2026

Cover Download Context Page Download
Dergi Kapağı
Social Sciences

Serial Number : 3C
ISSN No. : 1308-7444
Release Interval (in a Year) : 4 Issues

İSLAMİ GEOMETRİK DESENLERİN YAPAY ZEKA MODELLERİ TARAFINDAN ALGILANMASI: CHATGPT VE GEMİNİ KIYASLAMASI

ŞAMIL CAN GÜDER 1

Bu çalışma, ChatGPT (GPT-4o) ve Gemini (1.5 Pro) modellerinin İslami Geometrik Desenleri (İGD) algılama ve yorumlama kapasitelerini kültürel, tarihsel ve mimari bağlamda kıyaslamayı amaçlamaktadır. Araştırmada yöntem olarak, geleneksel kurallara uygun tasarlanan 12 özgün desen; sıfır-örneklem, düşünce zinciri ve mekansal bağlam istemlerini içeren üç kademeli bir deneysel protokolle test edilmiştir. Bulgular, yerleşik çok kipli mimariye sahip Gemini modelinin, bu çalışmada kullanılan veri seti üzerindeki desenleri saptamada yüzde 100 başarı gösterdiğini; metinsel muhakeme odaklı ChatGPT’nin ise desenleri tekil analiz etmek yerine kanıtlanamayan çıkarımlar üreterek tam başarıda yüzde 0’da kaldığını ortaya koymuştur. ChatGPT’nin özellikle Batı dışı motiflerde kültürel düzleşme yaşadığı ve gerçekliğe dayanmayan geometrik veriler ürettiği saptanmıştır. Sonuç olarak, bu deney kapsamında belirli sınırlılıklar gözlenmiş olup, yapay zeka modellerinin kültürel miras analizlerinde tam anlamıyla güvenilir hale gelmesi için eğitim veri setlerindeki Batı merkezli dengesizliğin bölgesel ve teknik metriklerle giderilmesi bilimsel bir zorunluluktur.

Keywords
İslami Geometrik Desenler, Yapay Zeka, Üretken Tasarım, Bilgisayarlı Görü, Kültürel Düzleşme,

PERCEPTION OF ISLAMIC GEOMETRIC PATTERNS BY ARTIFICIAL INTELLIGENCE MODELS: A COMPARISON OF CHATGPT AND GEMINI

ŞAMIL CAN GÜDER 1

This study aims to compare the perception and interpretation capacities of ChatGPT (GPT-4o) and Gemini (1.5 Pro) models regarding Islamic Geometric Patterns (IGP) within cultural, historical, and architectural contexts. As a methodology, 12 original designs following traditional rules were tested using a three-stage experimental protocol including zero-shot, chain-of-thought, and spatial context prompts. Findings revealed that the Gemini model, with its native multi-modal architecture, achieved 100% success in identifying the patterns within the dataset used in this study, while ChatGPT, focused on textual reasoning, achieved 0% full success by generating unverified inferences instead of individual analysis. It was determined that ChatGPT suffers from cultural flattening and produces geometric data not based on reality, particularly in non-Western motifs. Consequently, specific limitations were observed in this experiment, and to make AI models reliable in cultural heritage analysis, it is a scientific necessity to address the Western-centric imbalance in training datasets through regional and technical metrics.

Keywords
Islamic Geometric Patterns, Artificial Intelligence, Generative Design, Computer Vision, Cultural Flattening,

Details
   

Authors

ŞAMIL CAN GÜDER (1) (Corresponding Author)
NİŞANTAŞI ÜNİVERSİTESİ
samilcan.guder@nisantasi.edu.tr | 0000-0002-5335-9098

Supporting Institution

 :

Project Number

 :

Thanks

 :
References
1. Al-Ayyoub, M., vd., (2022). Classification of Islamic Geometric Patterns based on Machine Learning Techniques. ResearchGate.

2. Alghamdi, R., (2026). KingdomGlimpses: Evaluating Saudi Cultural Representation Through Text-to-Image Models. IEEE Xplore.

3. Alotaibi, S., (2025). Identification of Features of Architectural Heritage Using Deep Learning Techniques. HRPub.

4. Al-Saif, N., (2025). Using transfer learning to determine the type of mathematical fractals image of Islamic geometric patterns. R Discovery.

5. Aly, N. and Eltarabily, S., (2025). Heritage-Aware Generative AI Workflow for Islamic Geometry in Interiors. MDPI.

6. Andersson, L., (2025). Evaluating the performance of GPT-4o and Gemini 1.5 Pro on algorithms and coding problems. DiVA.

7. Ardatürk, A.Ş. (2022). Tasarımcı Zihninin Bir Yansıması Olarak; "Yapay Zeka". Online Journal of Art and Design, 10(4), 283-300.

8. Chen, Y., vd., (2025). AI-Enabled Cultural Experiences: A Comparison of Narrative Creation Across Different AI Models. MDPI.

9. Data Science Team., (2025). Prompting Vision Language Models. Exploring techniques to prompt VLMs. Medium.

10. Djibril, M. ve Thami, R.O., (2008). Islamic geometrical patterns indexing and classification using discrete symmetry groups. Semantic Scholar.

11. Eloy, S., vd., (2023). Mosque Typo-Morphological Classification for Pattern Recognition Using Shape Grammar Theory. MDPI.

12. Garcia, M., vd., (2024). Have Large Vision-Language Models Mastered Art History?. arXiv.

13. Google Research., (2024). Benchmarking Vision Language Models for Cultural Understanding. arXiv

14. Hu, X., vd., (2025). Advancing vision-language models with spatial-context prompt tuning. Taylor & Francis.

15. Hugging Face., (2026). Daily Papers: GPT-4o Context Limits. Hugging Face.

16. Khan, F., (2026). A new canvas of learning: Enhancing formal analysis skills in AP art history through AI-generated Islamic art. FES Journals.

17. Lee, C., vd., (2026). VULCA-Bench: A Multicultural Vision-Language Benchmark for Evaluating Cultural Understanding. arXiv.

18. Li, J., vd., (2025). A Survey of State of the Art Large Vision Language Models: Alignment, Benchmark, Evaluations. arXiv.

19. Moustapha, H. ve Krishnamurti, R., (2001). Application-based principles of islamic geometric patterns; state-of-the-art, and future trends. OUCI.

20. Neontri., (2026). ChatGPT vs. Gemini: Which AI Listens to You Better?. Neontri Blog.

21. Reid, M., Savinov, N., Teplyashin, D., Co, D., Yang, A., Antropov, K., et al. (2024). Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context. arXiv preprint arXiv:2403.05530.

22. Reisig, L., (2016). Examining Islam Through Art: Islamic Geometric Patterns. Harvard Blogs.

23. Roberts, D., (2026). Artificial Intelligence and Misinformation in Art. ResearchGate.

24. Smith, A., (2026). Digital heritage and fashion: preserving cultural identity through AI. Oxford Academic.

25. Vayvay, B.G., & Bove, G., (2023). The Semiology of Cult-Inspired Souvenirs: Museum Artefacts and the Ephesian Artemis. Journal of Archaeology & Art (Arkeoloji ve Sanat), 174, 245-254.

26. Wang, T., (2026). ExAnte: A Benchmark for Ex-Ante Inference in Large Language Models. ACL Anthology.

27. Youssef, M., (2024). Artificial Intelligence Islamic Architecture (AIIA): What Is Islamic Architecture in the Age of AI?. MDPI.

28. Zhang, L., vd., (2026). Appear2Meaning: A Cross-Cultural Benchmark for Structured Cultural Metadata Inference. arXiv.