Towards Interactive AI-assisted Material Selection for Sustainable Building Design

Abstract

We present an AI-assisted workflow that supports architects in designing wall assemblies using sustainable materials. Material selection in architecture is a complex process involving multiple data points and trade-offs across environmental performance, cost, and constructability. Making this process more efficient is essential for encouraging sustainable design practices. Our approach uses artificial intelligence and large language models to streamline aspects of material analysis and information management. The workflow integrates into standard architectural practice by translating wall assembly sketches into graph representations that reflect components and their relationships. Through an interactive interface with graph visualization, architects can explore material options, review properties and substitute components in line with their design intent. We contribute a prototype workflow and report findings from a preliminary study on the integration of AI tools in early design stages. The study highlights benefits such as reduced decision effort, increased confidence, and improved access to material information.