Study on use of Text Based AI Tool in Construction Industry Contracts

Abstract/Description

This study aimed at carrying out exploratory research on the conversational AI tools defined in literature for contract text drafting and review. The term conversation AI refers to any text based intelligent system developed to imitate or simulate human-machine interaction. In this project, the literature review was carried out by retrieving research papers from Google Scholar, Scopus and by directly approaching the authors of research papers having restricted access on ASCE library. Keyword search method with Boolean operators was used. For development of NLP model, Rasa, an open-source AI platform for building chatbots and assistants was used. The model was tested and trained on three clauses obtained from fourth edition of FIDIC General Conditions of Contract version 1987 (reprinted in 1990) which is referred by the Pakistan Engineering Council in its Standard Bidding Document on its website. The results of the model were reviewed manually, confusion matrix was generated by Rasa on use of relevant commands. It was observed that because of legal nature of the text, there was limitation in variation that could easily capture the core message of the text for training of the model. The training of conversational AI model becomes more difficult with increased length of the text as generating variations becomes more tedious task. Since, the 4 application of this model is for local construction industry, and there was a limited scope of model development, comparison of conversational AI models is not easily generalizable. The shift of Pakistan’s construction industry to use of AI tools in contract administration and management is inevitable as the world moves to adopt such technological advancements, the same might take more time than in the developed countries. However, there is need to assess the readiness of the industry and identification of factors that would stimulate as well as hinder the adoption of such technologies in Pakistan, which has been proposed as future area of research. The research can be expanded to include private sector contracts by forming a database and allowing automated contract clause formulation upon user query..

Track

Accounting & Law

Session Number/Theme

1D: Accounting & Law

Session Chair

Dr. Danish Godil ; Dr. Asif Jaffar

Start Date/Time

30-5-2024 1:50 PM

End Date/Time

30-5-2024 3:20 PM

Location

MCS – 11 AMAN CED Building

This document is currently not available here.

Share

COinS
 
May 30th, 1:50 PM May 30th, 3:20 PM

Study on use of Text Based AI Tool in Construction Industry Contracts

MCS – 11 AMAN CED Building

This study aimed at carrying out exploratory research on the conversational AI tools defined in literature for contract text drafting and review. The term conversation AI refers to any text based intelligent system developed to imitate or simulate human-machine interaction. In this project, the literature review was carried out by retrieving research papers from Google Scholar, Scopus and by directly approaching the authors of research papers having restricted access on ASCE library. Keyword search method with Boolean operators was used. For development of NLP model, Rasa, an open-source AI platform for building chatbots and assistants was used. The model was tested and trained on three clauses obtained from fourth edition of FIDIC General Conditions of Contract version 1987 (reprinted in 1990) which is referred by the Pakistan Engineering Council in its Standard Bidding Document on its website. The results of the model were reviewed manually, confusion matrix was generated by Rasa on use of relevant commands. It was observed that because of legal nature of the text, there was limitation in variation that could easily capture the core message of the text for training of the model. The training of conversational AI model becomes more difficult with increased length of the text as generating variations becomes more tedious task. Since, the 4 application of this model is for local construction industry, and there was a limited scope of model development, comparison of conversational AI models is not easily generalizable. The shift of Pakistan’s construction industry to use of AI tools in contract administration and management is inevitable as the world moves to adopt such technological advancements, the same might take more time than in the developed countries. However, there is need to assess the readiness of the industry and identification of factors that would stimulate as well as hinder the adoption of such technologies in Pakistan, which has been proposed as future area of research. The research can be expanded to include private sector contracts by forming a database and allowing automated contract clause formulation upon user query..