Life Cycle Assessment for the Unconventional Construction Materials in Collaboration with a Large Language Model

Abstract

In this paper, developing an online tool for the Life Cycle Assessment (LCA) of unconventional construction materials in collaboration with Large Language Models (LLMs) is proposed. The LCA provides information on the environmental impact of a product throughout its entire life cycle, from the extraction of raw materials to disposal or recycling. The LLMs are neural network architectures, typically utilizing variants of recurrent neural networks such as the transformer, which are trained on large bodies of textual data using techniques such as pre-training and fine-tuning. This study focuses on the use of bacterial cellulose composites as a biobased unconventional construction material. The methodology of developing an LLM-aided LCA tool is divided into five stages: Defining the functional unit; identifying the life cycle stages; collecting environmental and social impact data; interpreting and evaluating; developing a web-based tool. The results of this study have shown that the designers can incorporate sustainable thinking in the design process by using LLMs integrated to LCA, ultimately contributing to a more sustainable future against the impacts of the Anthropocene. Overall, the outcomes demonstrated the value of human-computer interaction (HCI) as a tool for exploring new possibilities with biobased materials and for inspiring designers to reconsider the material evaluation in their work. Future studies can delve into the integration of this tool into building information modeling software or computational design software in order to perform LCA for 3D structures. Different scales of such applications in design practices, such as fashion design, product design or service design can also be conducted by questioning how LCA can be combined with LLMs to leverage novel sustainable design solutions.