Those working in the architecture, engineering, and construction (AEC) domain often get bogged down in day-to-day paperwork and verbally communicated instructions. Natural language processing (NLP)-based artificial intelligence (AI) would certainly help. The authors call this approach conversational AI (CAI) for AEC. This paper surveys the literature on CAI for AEC and will interest researchers in the field.

After a brief introduction to CAI and AEC, the authors give a general review of the available CAI literature, highlighting progress in various application areas. The third section gives an overview of CAI categories and system architecture. They point out parameters for CAI categorization: modularity, channels, approach, communication, objective, and integration methods, while implementation approaches can be rule based or standard data driven. The choice of system architecture may depend on the dialog management system and automatic speech recognition methods (such as algorithm-based methods using Gaussian mixture models, hidden Markov models, mel-frequency cepstral coefficients (MFCCs), n-gram language models, and natural language understanding).

Their research methodology included search on Scopus and the web; validation using IEEE Xplore, ACM, and ScienceDirect; followed by the selection process. The selected publications fell under the construction industry, the Construction Advancement Institute (CAl), the AI category, or a combination of these. The selected 42 publications were then tabulated using an Excel spreadsheet with columns for author, year, title, publication type, aim, methodology, and domain area. Using this data, the authors show that it is the early days of CAI in the AEC industry; research and opportunities exist in all phases of construction, including design, post-construction, and value-adding.