Explainable Artificial Intelligence in Generative Design For Modular Construction

Authors

  • Sanaz Zarghami Department of Architecture and Built Environment, Northumbria University, United Kingdom
  • Hanieh Kouchaki Department of Architecture and Urbanism, Tabriz Islamic Art University, Iran
  • Longzhi Yang Department of Computer Information Sciences, Northumbria University, United Kingdom
  • Pablo Martinez 1 Department of Architecture and Built Environment, Northumbria University, United Kingdom

Keywords:

Explainable AI; Generative design; Modular construction

Abstract

As artificial intelligence rapidly advances, its growing complexity enables more sophisticated applications across sectors, including modular construction. However, the opaque nature of algorithms, such as generative AI, reduces human interpretability and trust. While providing benefits like enhanced efficiency, the black-box processes of generative design hamper adoption. Explainable AI can elucidate how AI algorithms generate outputs, thereby improving understanding and confidence. Despite explainable AI’s potential, construction has given it limited focus. This research systematically reviews the application of explainable AI in generative design in construction, with an aim to allay risks and enable wider utilization of these emerging technologies for improved engineering design across modular buildings. This research proposes a categorization system for 'Explainable Generative Artificial Intelligence Design,' which organizes and differentiates various GENAI models, while also offering approaches to improve their ability to provide explanations and increase transparency.

Downloads

Published

2024-06-26

Conference Proceedings Volume

Section

Academic Papers