Assessing the role of Al in advancing construction sector industrial symbiosis research: A comparative study of leading digital assistants

Abstract

The integration of Artificial Intelligence (AI) in the construction sector has opened new avenues for advancing Industrial Symbiosis (IS) research. However, existing literature lacks a comprehensive comparison of how leading AI digital assistants contribute to this field. This study addresses this gap by examining the performance of four prominent AI models, Gemini, CoPilot, ChatGPT-Classic, and ChatGPT-Advanced in generating responses related to IS opportunities in construction industry. The methodology involves a two-stage analysis: first, questions related to IS concepts and practices are posed to each AI model to test their response reproducibility, measured using BLEU, METEOR, and Cosine Similarity scores. This is followed by human expert evaluations to validate the quality of the responses. In the second stage, the models are tasked with defining the European Waste Catalogue (EWC) codes and Statistical Classification of Economic Activities in the European Community (NACE) sector classifications associated with the selected waste materials, followed by identifying potential IS opportunities. Key findings reveal significant variability in the models' capabilities. ChatGPT models consistently demonstrate higher semantic alignment with expert evaluations in both the general questions and IS opportunity identification. In contrast, CoPilot shows strengths in syntactic accuracy but sometimes lacks depth in contextual understanding. The study also identifies that while some AI models are adept at defining waste codes and sector classifications, their ability to identify practical IS opportunities varies. These insights underscore the need for an integrated approach, combining AI-generated data with human expertise, to fully exploit IS potential in construction. This study not only sheds light on the current state of AI in IS identification but also provides a framework for evaluating AI models in similar contexts. Future studies should focus on enhancing AI models' contextual understanding and broadening their applications to promote sustainable industrial practices across various sectors.