Investigating the Role of Artificial Intelligence Technologies in the Construction Industry Using a Delphi-ANP-TOPSIS Hybrid MCDM Concept under a Fuzzy Environment

Abstract

The construction business is always changing, and with the introduction of artificial intelligence (AI) technology it is undergoing substantial modifications in a variety of areas. The purpose of this research paper is to investigate the function of AI tools in the construction industry using a hybrid multi-criteria decision-making (MCDM) framework based on the Delphi method, analytic network process (ANP), and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) under a fuzzy scenario. The ANP framework offers a systematic approach to quantifying the relative importance of AI technologies based on expert opinions gathered during the Delphi process, whereas the fuzzy TOPSIS methodology is used to rank and select the most appropriate AI technologies for the construction industry. The final results from the ANP revealed that the technological factors are the most crucial, followed by the environmental factors, which highly influence the AI environment. In addition, TOPSIS identified robotics and automation as the best AI alternative among the three options, followed by building information modeling (BIM), whereas computer vision was the least preferred among the list. The proposed hybrid MCDM framework enables a comprehensive evaluation and selection process that takes into account the interdependencies between AI technologies and uncertainties in decision-making.

1. Introduction

The construction industry, traditionally known for its reliance on human labor and conventional practices, has experienced a significant transformation in recent years with the introduction of AI technologies. AI, with its ability to process vast amounts of data, analyze complex patterns, and make informed decisions, has revolutionized various aspects of the construction industry [1]. This paper aims to explore the role of AI technologies in construction, examining their impact on project planning, design, construction processes, and overall industry performance. In the past, the construction industry has faced numerous challenges, such as inefficiencies, cost overruns, and safety hazards. However, the adoption of innovation and technology has become crucial in addressing these issues. Artificial intelligence, as a cutting-edge technology, offers immense potential to enhance efficiency, accuracy, and safety in construction projects. AI encompasses various concepts and components, including machine learning, deep learning, natural language processing, and computer vision, all of which find applications in the construction sector [2]. For instance, AI algorithms can assist in site selection and feasibility analysis by considering multiple factors and data points. Design tools powered by AI can generate optimized plans and layouts, while virtual and augmented reality applications enable stakeholders to visualize construction projects before their actual implementation.

The impact of AI extends beyond the planning and design phases to the construction processes themselves. Robotics and automation technologies enable the use of intelligent equipment and machinery, reducing manual labor and increasing productivity. Predictive analytics play a vital role in resource management and scheduling, ensuring efficient utilization of materials, equipment, and labor. Safety and risk management have also been enhanced through AI technologies [3]. Real-time monitoring systems equipped with AI algorithms can detect safety violations and provide early warnings to prevent accidents and injuries. Additionally, predictive analytics can help identify potential risks and enable proactive mitigation measures, thereby minimizing construction-related hazards. The implementation of AI in construction offers numerous benefits. Improved efficiency and productivity result from reduced project timelines, optimized workflows, and error minimization through AI-driven quality control. Cost reduction and budget optimization are achieved through AI-based cost estimation, material usage optimization, and enhanced supply-chain management. Furthermore, AI enables quality enhancement and innovation by providing design tools for creative and optimized solutions and integrating with building information modeling (BIM) for improved collaboration and coordination among project stakeholders. However, there are challenges and limitations to consider in the adoption of AI technologies in construction [4]. Data availability and quality pose significant hurdles, as the industry historically lacks comprehensive and standardized data. Resistance to change and limited awareness among industry professionals can impede the widespread adoption of AI. Ethical considerations, including concerns about job displacement, also need to be addressed as AI increasingly becomes part of construction processes.

Looking ahead, the future of AI in construction holds promising possibilities. Emerging trends and advancements in AI technologies continue to shape the industry, driving further innovation and efficiency gains. However, careful consideration must be given to the potential impact on workforce roles and skills, ensuring that AI is implemented responsibly and sustainably within the industry. The role of artificial intelligence technologies in the construction industry is transformative [5]. From project planning and design to construction processes and safety management, AI offers opportunities to enhance efficiency, reduce costs, improve safety, and foster innovation. Although challenges exist, embracing AI’s potential and addressing ethical considerations will help ensure a successful and progressive integration of AI in the construction sector.

1.1. Motivations of Using the Hybrid MCDM Concept under a Fuzzy Environment

The Delphi-ANP-TOPSIS Hybrid MCDM approach was chosen for its ability to handle complicated decision-making situations including numerous criteria and uncertainties in a fuzzy environment. Let us break down the concept to better understand its motive.

• Delphi method: The Delphi method is a structured strategy for getting feedback from a large number of experts or stakeholders. It seeks to achieve consensus or convergence of opinions by gathering and improving expert assessments iteratively. The Delphi approach is useful in establishing valid and impartial criteria weights and preferences in the context of MCDM.
• ANP: The ANP is a decision-making technique that extends the analytic hierarchy process (AHP) to account for criteria interdependence. The ANP provides for the evaluation of both the influence and dependent relationships among the criteria, resulting in a more thorough study. The ANP aids in the capture of complicated linkages and interactions among criteria, which is useful in decision-making processes.
• TOPSIS: TOPSIS is a popular MCDM method for ranking options based on their resemblance to the ideal solution. It computes the relative distances of each alternative between the ideal and anti-ideal solutions. To establish the overall rankings, TOPSIS considers both the good and negative aspects of the criteria.
• Fuzzy environment: There is uncertainty or ambiguity in the data or criteria utilized in many real-world decision-making circumstances. Fuzzy logic, which allows for degrees of membership or partial truth, provides a framework for dealing with ambiguity. It enables the representation of ambiguous or imprecise data, which is especially valuable when dealing with subjective judgments or language assessments.

The goal of merging these strategies in a hybrid framework is to capitalize on their own strengths while addressing the limitations of individual methods. Expert opinions can be pooled and consensus attained by utilizing the Delphi technique. The ANP allows for the modeling of criteria interdependencies, encapsulating the complexities of decision-making problems. TOPSIS offers a systematic method for ranking options based on their proximity to the optimal answer. Finally, by including a fuzzy environment, the model can deal with uncertainties and imprecise information that are common in real-world decision-making scenarios. Hence, the goal of using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment is to provide a robust and comprehensive decision-making framework that accounts for complex criteria relationships, expert opinions, and uncertainties, ultimately leading to more informed and reliable decisions.

1.2. Significance of AI Technologies in the Construction Industry

The significance of AI technologies in the construction industry cannot be overstated. The introduction of AI has revolutionized traditional construction practices and brought about a multitude of benefits and advancements [6,7]. Here are some key aspects highlighting the significance of AI in the construction industry:

• Improved efficiency and productivity: AI technologies optimize various processes involved in construction projects, leading to increased efficiency and productivity [8]. By automating repetitive tasks and leveraging predictive analytics, AI streamlines workflows, minimizes errors, and reduces project timelines. This enables construction companies to complete projects faster and with greater precision, ultimately improving their overall operational efficiency.
• Enhanced safety and risk management: Safety is a critical concern in the construction industry, and AI technologies play a pivotal role in mitigating risks and ensuring the wellbeing of workers. Real-time monitoring systems, equipped with AI algorithms, can detect and alert stakeholders about potential safety hazards, allowing for prompt intervention [9]. Predictive analytics can also identify high-risk areas or situations in advance, enabling proactive measures to prevent accidents and injuries.
• Cost reduction and budget optimization: AI technologies provide construction companies with valuable tools and insights for cost reduction and budget optimization. AI-based cost estimation algorithms consider various parameters, historical data, and market trends to generate accurate cost projections [10,11]. This helps construction firms in developing realistic budgets and reducing the likelihood of cost overruns. Additionally, AI assists in optimizing resource allocation, material usage, and supply-chain management, resulting in substantial cost savings.
• Improved design and planning: AI technologies offer advanced design and planning capabilities that enhance the quality and efficiency of construction projects. AI-powered design tools can generate optimized plans and layouts based on specific requirements and constraints [12]. Virtual and augmented reality applications allow stakeholders to visualize projects in a realistic manner, facilitating better communication, collaboration, and decision-making during the design and planning stages.
• Quality enhancement and innovation: AI technologies facilitate quality enhancement and foster innovation in the construction industry. By integrating AI with building information modeling (BIM), construction professionals can optimize designs, identify clashes, and simulate construction processes before implementation, leading to improved quality and reduced rework [2,6]. AI-powered algorithms can also analyze historical data to identify patterns and generate innovative solutions for complex construction challenges.
• Data-driven decision-making: The construction industry generates vast amounts of data throughout the project lifecycle. AI technologies enable construction companies to harness the power of these data and derive actionable insights. Machine learning algorithms can analyze large datasets to identify patterns, trends, and correlations, enabling data-driven decision-making [4,9]. This empowers construction professionals to make informed choices regarding project planning, risk management, resource allocation, and other critical aspects.
• Sustainable construction practices: AI technologies contribute to the adoption of sustainable construction practices. By optimizing resource usage, energy consumption, and waste management, AI helps reduce the industry’s environmental impact [1]. AI-enabled sensors and systems can monitor energy usage in buildings and suggest energy-saving measures. Additionally, AI algorithms can analyze data related to building materials and suggest eco-friendly alternatives, promoting sustainable construction practices.

AI technologies have significantly transformed the construction industry, offering numerous advantages and opportunities for improvement. From improving efficiency and productivity to enhancing safety, reducing costs, and driving innovation, AI plays a pivotal role in shaping the future of construction [8,10]. By harnessing the power of AI, construction companies can unlock new levels of performance, competitiveness, and sustainability.

1.3. Strengths and Weaknesses of the Fuzzy Delphi-ANP-TOPSIS Hybrid MCDM Model

The Delphi-ANP-TOPSIS hybrid approach offers several benefits and has some limitations. Let us explore some of the benefits and weaknesses of the hybrid MCDM concept.

1.3.1. Strengths

Here are some of the benefits of the established hybrid model. The hybrid system consists of three MCDM models, namely, the Delphi method, ANP, and TOPSIS, which were chosen due to having some significant strengths over other MCDM tools.

• The Delphi-ANP-TOPSIS technique allows for the integration of numerous expert viewpoints and perspectives through the Delphi technique. It improves the impartiality and comprehensiveness of the decision-making process by requesting and synthesizing feedback from a varied collection of specialists. In this case, with the help of the Delphi managerial technique, the experts were able to finalize the most suitable parameters and eliminate the irrelevant ones.
• In the hybrid MCDM framework, the ANP helps in investigating the complicated interdependencies among the criteria. It allows decision-makers to collect and assess the interactive feedback loops among various components, resulting in a more accurate and comprehensive decision model. Here, the ANP contributed in the establishment of a hierarchical network depicting the relationships among the criteria and alternatives. The ANP also helps in evaluating the criteria weightages.
• The TOPSIS method allows for the systematic evaluation and ranking of available alternatives. It takes into account both the positive and negative elements of decision criteria, resulting in a more comprehensive evaluation of alternatives. TOPSIS assisted in the ranking of three AI alternatives in this investigation, allowing the experts to choose the superior one for the construction industry.
• The employment of the fuzzy concept helps to address uncertainties and ambiguity in decision-making. Fuzzy logic and linguistic variables help decision-makers to deal with the inherent uncertainties in real-world problems by simulating inaccurate and subjective assessments.
• The Delphi-ANP-TOPSIS idea provides an organized and methodical decision-making framework. It guides decision-makers through the process, ensuring that appropriate criteria are taken into account, expert viewpoints are incorporated, and interdependencies are treated properly. This methodical approach encourages consistency and transparency in decision-making.

1.3.2. Weaknesses

Apart from their benefits, all MCDM tools also have some limitations that should not be overlooked completely. This section highlights some of the drawbacks of the hybrid MCDM model.

• Due to the engagement of several experts and repetitive rounds of information gathering and consensus building, the Delphi-ANP-TOPSIS approach can be time-consuming and costly. To collect expert perspectives and create a consensus, the Delphi procedure necessitates a significant amount of time and effort, which may not be possible in time-sensitive decision circumstances.
• The reliance on expert opinions and judgments in the data collection phase introduces subjectivity and the potential for bias. The quality and reliability of expert inputs may vary, and individual biases can influence the outcomes of the decision-making process.
• The integration of different techniques like fuzzy Delphi, ANP, and TOPSIS within the hybrid framework necessitates some level of competence in each method. Implementing the hybrid concept may necessitate specific knowledge and abilities, making it more difficult for decision-makers who are unfamiliar with these methods.
• The Delphi-ANP-TOPSIS approach may face limitations when applied to large-scale decision problems or those involving numerous criteria and alternatives. As the complexity and scale of the decision context increase, the implementation and management of the Delphi-ANP-TOPSIS process can become more challenging and time-consuming.
• The ANP relies on pairwise comparisons to determine the relative weights of criteria. While this procedure offers an organized approach, it might still be subjective and lack transparency. The transparency of the weight assignment process is important for decision-makers and stakeholders to understand and trust the decision outcomes.

Although the established hybrid model has some limitations, the strengths discussed in the previous section are so compelling that it has the ability to overcome its weaknesses. The authors were motivated by the positive features of the hybrid model and decided to adopt it for the present analysis.

1.4. Research Objective and Problem Statement

The objective of this research paper is to investigate and analyze the role of AI technologies in the construction industry using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment. The research aims to provide a comprehensive understanding of how AI technologies can be effectively utilized in the construction sector, considering the complex and uncertain nature of construction projects. The construction industry faces numerous challenges, such as optimizing project planning, design, resource allocation, and risk management. The integration of AI technologies has the potential to address these challenges and improve overall industry performance. However, there is a lack of research that systematically evaluates and quantifies the role of AI technologies in the construction industry, considering the uncertainties and fuzziness associated with decision-making in construction projects.

To fill this research gap, this study aims to investigate the role of AI technologies in the construction industry by employing a Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment. The research seeks to address the following key questions:

i.

What are the critical criteria and sub-criteria that need to be considered when evaluating the performance and impact of AI technologies in the construction industry?

ii.

How can the Delphi-ANP-TOPSIS hybrid MCDM concept be applied to assess the role of AI technologies in the construction industry, accounting for the uncertain and fuzzy nature of decision-making?

iii.

What are the potential benefits, challenges, and limitations associated with the integration of AI technologies in the construction industry, considering the fuzzy environment?

iv.

How can the findings of this research contribute to enhancing decision-making processes and improving the adoption and implementation of AI technologies in the construction industry?

By addressing these research questions, this study aims to provide valuable insights into the effective utilization of AI technologies in the construction industry and contribute to the development of decision-making frameworks for evaluating AI-driven solutions under a fuzzy environment.

2. Literature Review

The construction industry is undergoing a transformative phase with the integration of artificial intelligence (AI) technologies. AI offers the potential to revolutionize various aspects of the industry, including project planning, design, construction processes, and safety management [13,14]. This literature review aims to provide an overview of key studies and research conducted on AI technologies in the construction industry, highlighting their applications, benefits, and challenges. Several brainstorming sessions were held with panel experts who have extensive knowledge and expertise on the topic. The Scopus and Web of Science (WoS) databases were mostly used to gain access to a large number of research articles. This article primarily follows and refers to high-impact-factor (IF) peer-reviewed international journals indexed in Scopus and WoS. A search for related keywords yielded 36,000 (approx.) published articles in the ScienceDirect database. The authors mostly used databases from internationally renowned publishers such as Elsevier, Springer, Wiley, Emerald, Sage, Taylor & Francis, etc. The authors sorted around 450 articles from a vast list of published articles based on the key theme of the article. Following that, the authors removed some of the puzzled and irrelevant papers from the list. A series of filters were also used to further narrow the list. Finally, around 150 publications were sifted and thoroughly researched in order to discover the critical parameters that most influence AI technologies in the construction business. The authors found 13 parameters, as listed in

Table 1. List of selected parameters, divided into three broad categories.

Main Criteria (MC) Category	MC Indicators	Sub-Criteria (SC) Category	SC Indicators
Technology criteria	TC	Complexity	CR1
		Aesthetics	CR2
		Value of data and algorithms	CR3
		Advancements in innovations for the traditional problems	CR4
Organization criteria	OC	Government/management	CR5
		Cost/sufficient budget	CR6
		Employee workforce	CR7
		Information exchange and communication/interoperability	CR8
		Risk-taking ability	CR9
Environment criteria	EC	Upstream and downstream policy/laws	CR10
		Trust between different companies/copyright/ownership	CR11
		Social impacts	CR12
		Regulatory measures	CR13

Source: Interpreted by the expert team members.

, that were consistent themes throughout the ongoing analysis, because they were considered by several researchers multiple times in their past studies. Below are some of the published studies that demonstrate the utility of various MCDM tools and the impact of AI technologies in the construction industry.

2.1. Prior Studies Unveiling the Influence of AI in the Construction Industry

AI technologies play a crucial role in project planning and design. AI has been gaining traction as a tool to enhance efficiency, safety, and project outcomes in the construction industry. However, the complexity of AI systems can also present obstacles to their implementation and utilization in the field [9]. Several researchers have emphasized that the complexity of AI systems can lead to a lack of transparency and interpretability, making it difficult for construction professionals to understand and trust the decisions made by the AI, hindering the adoption and implementation of AI in the industry [4,5]. Sacks et al. [15] discussed the utilization of AI algorithms for site selection and feasibility analysis, enabling informed decision-making. AI-powered design tools, as explored by Na et al. [16], generate optimized plans and layouts, improving efficiency and accuracy. Virtual and augmented reality applications, as studied by Pan and Zhang [2], facilitate project visualization and enhance stakeholder collaboration. AI technologies have significant implications for construction processes. Robotics and automation, as highlighted by Afzal et al. [7], can reduce manual labor, enhance productivity, and improve safety. Predictive analytics, examined by Oprach et al. [11], can enable resource management and scheduling optimization, resulting in efficient resource allocation. Additionally, intelligent equipment and machinery, as discussed by Xie et al. [14], can enhance construction productivity and quality. Ensuring safety and mitigating risks are critical in the construction industry. AI technologies offer innovative solutions in this domain. Real-time monitoring systems integrated with AI algorithms, as explored by Aziz et al. [17], can detect safety violations and enable proactive measures. Predictive analytics, as investigated by Turner et al. [18], can assist in identifying and mitigating potential hazards. AI-driven risk assessment and decision-support systems, as studied by Bang and Andersen [19], can improve safety management processes.

Apart from these, many studies have pointed out that the complexity of AI systems can also result in issues with data quality and availability, as well as a lack of understanding among construction professionals on how to properly use and maintain the systems. Additionally, Aljawder and Al-Karaghouli [10] also highlighted how the complexity of AI systems can make it difficult for construction professionals to properly evaluate the performance and effectiveness of the systems, which can impede decision-making and progress in implementing AI in the construction industry. Therefore, it is crucial for researchers and practitioners in the construction industry to take into account the complexity of AI systems and address these challenges to effectively implement and utilize AI in the construction industry.

2.2. Benefits of AI Implementation

The integration of AI technologies in the construction industry leads to improved efficiency and productivity. Egwim et al. [3] demonstrated that AI-driven automation reduces project duration and increases output. Wang et al. [20] discussed the challenges of adopting AI and the IoT by analyzing the causal relationships. Moreover, AI-driven quality control, as examined by Roslon [4], reduces errors and rework, further enhancing efficiency. AI technologies contribute to cost reduction and budget optimization in construction projects. Regona et al. [6] showed that AI-based cost estimation improves accuracy and aids in budget forecasting. Material usage optimization, as discussed by Zavadskas et al. [8], reduces waste and optimizes procurement processes. Additionally, AI-driven supply-chain management, as explored by Yaseen et al. [21], enhances efficiency and cost-effectiveness. AI technologies can play a crucial role in improving safety and mitigating risks. Oluleye et al. [9] demonstrated that AI-powered monitoring systems can enhance worker safety by providing real-time alerts and warnings. Predictive analytics, as discussed by Holzmann and Lechiara [22], can enable proactive risk identification and mitigation. This results in accident prevention and improved overall safety performance in construction projects. AI technologies foster quality enhancement and drive innovation in the construction industry. Heo et al. [23] highlighted that AI-powered design tools can enable creative and optimized solutions, improving the quality of construction projects. The integration of AI with building information modeling (BIM), as explored by Akinosho et al. [24], enhances collaboration and coordination among project stakeholders, leading to improved quality and innovation. AI algorithms, as discussed by Srivastava et al. [25], can analyze historical data to identify patterns and generate innovative solutions for complex construction challenges, promoting continuous improvement and innovation in the industry.

2.3. Identification of Key Challenges and Opportunities Associated with AI’s Adoption in the Construction Industry

The construction industry has witnessed a growing interest in the adoption of AI technologies to address various challenges and leverage opportunities for improved efficiency, productivity, and safety. However, the successful integration of AI in construction requires the identification and assessment of key challenges and opportunities. This literature review explores the adoption of AI in the construction industry, using MCDM techniques to identify these factors.

2.3.1. Challenges of AI Adoption in the Construction Industry

One of the key challenges in the adoption of AI is data quality and integration. The construction industry generates vast amounts of data from different sources, including project plans, sensor data, and historical records. Ensuring the quality, compatibility, and integration of these data is crucial for effective AI implementation [19]. Technical and technological barriers pose another challenge. The adoption of AI requires a robust technical infrastructure, including high-performance computing, reliable network connectivity, and interoperability between different AI applications. Overcoming these barriers is essential to harness the full potential of AI in construction [22,23]. Ethical and legal considerations are also important challenges to address. The use of AI in construction raises concerns related to privacy, security, and liability. It is necessary to establish guidelines and regulations to govern the responsible and ethical use of AI technologies in the industry [14].

2.3.2. Opportunities of AI’s Adoption in the Construction Industry

AI’s adoption in construction presents several opportunities for improvement. Process optimization and efficiency are among the key benefits. AI technologies can automate routine tasks, optimize resource allocation, and streamline workflows, leading to increased productivity, reduced costs, and improved project timelines [26]. Enhanced decision-making and risk management are also notable opportunities. AI-based analytics and predictive models can enable data-driven decision-making, risk assessment, and mitigation strategies. By analyzing historical data and real-time information, AI can support informed decision-making, leading to improved project outcomes. Safety and quality improvements are additional advantages of AI’s adoption. Computer vision and sensor technologies can enhance safety on construction sites by monitoring worker behavior, detecting hazards, and predicting potential accidents [27]. AI can also improve quality control processes, reducing errors and ensuring compliance with industry standards.

2.4. The Role of MCDM in the Adoption of AI Technologies in the Construction Industry

Numerous MCDM techniques have been applied in various domains within the construction industry. The analytic hierarchy process (AHP) has found applications in project selection, risk assessment, and supplier evaluation [28]. TOPSIS has been utilized in contractor selection, facility location, and project prioritization [29,30]. ELECTRE methods have been employed in facility management, supplier selection, and construction project prioritization, allowing decision-makers to consider preference thresholds. These MCDM techniques provide structured frameworks, ranking mechanisms, and preference considerations, contributing to informed decision-making processes in the construction industry. The use of MCDM techniques is valuable for identifying and evaluating the key challenges and opportunities associated with AI’s adoption in the construction industry. Approaches such as the AHP, fuzzy logic, and ELECTRE provide structured frameworks to assess criteria based on their relative importance [31]. MCDM enables decision-makers to systematically prioritize factors, facilitating informed decision-making in AI’s implementation. AI’s adoption in the construction industry offers significant opportunities for process optimization, enhanced decision-making, and improved safety and quality. However, challenges such as data quality, technical barriers, and ethical considerations must be addressed for successful integration. By utilizing MCDM techniques, decision-makers can systematically identify and assess these challenges and opportunities, guiding effective decision-making and supporting the successful adoption of AI technologies in the construction sector.

Tsai et al. [32] provided a systematic review focusing on the integration of MCDM techniques in sustainable construction decision-making. It explored how MCDM methods are used to assess and evaluate sustainability criteria and make informed decisions regarding green materials, energy-efficient technologies, and sustainable construction practices. A study by Zolfani et al. [33] provided an overview of various MCDM techniques and their applications in the construction project management of hotels. They discussed the use of techniques such as SWARA and COPRAS in construction decision-making processes. Matic et al. [34] examined the application of various MCDM techniques in construction project management for the proper evaluation of suppliers. They discussed the use of FUCOM, COPRAS, ARAS, WASPAS, MABAC, and SAW in areas such as project selection, risk assessment, supplier selection, and sustainable construction. The study provided a critical evaluation of the techniques and identified future research directions. Chatterjee et al. [35] reviewed the application of the ANP and MABAC MCDM techniques in a construction project for risk management and sustainability evaluation. The study discussed the strengths and weaknesses of various MCDM methods and provided insights into their implementation challenges. Rezakhani [36] developed a fuzzy embedded rational MCDM model for the effective selection of risk factors in construction projects. The study discussed the advantages, limitations, and suitability of various MCDM methods in construction decision-making.

The Delphi method, ANP, and TOPSIS are well-established decision-making methodologies that have been widely used across various disciplines. The Delphi method is a structured approach for obtaining expert consensus on a particular subject through a series of iterative questionnaires or surveys. It has been employed in diverse fields, including technology adoption, policymaking, and risk assessment. The Delphi method’s effectiveness in capturing and synthesizing expert opinions while maintaining anonymity has been well documented. A seminal study by Kulejewski and Rosłon [37] provides an in-depth exploration of the Delphi method’s techniques and applications, highlighting its ability to reduce biases and enhance the reliability of results. The ANP is an extension of the AHP that addresses complex decision problems by considering interdependencies among criteria. The ANP has found widespread use in fields such as project selection, supply-chain management, and sustainability assessment. Past researchers have contributed significantly to the understanding and application of the ANP. Works on the theory and applications of the ANP provide a comprehensive overview of the ANP, highlighting its ability to model complex decision problems and capture interdependencies among criteria. TOPSIS is another MCDM method used for ranking alternatives based on their proximity to an ideal solution. It compares alternatives to both the best and worst solutions, considering multiple criteria simultaneously. TOPSIS has been successfully applied in various domains, including supplier selection, facility location, and project prioritization. Being one of the oldest MCDM techniques, numerous researchers have made significant contributions to the development and application of TOPSIS. Several prior works provide an extensive overview of TOPSIS and its variations, showcasing its simplicity and effectiveness in ranking alternatives based on multiple criteria.

2.5. Fuzzy-Logic-Based Decision-Making Analysis in the Construction Industry

Fuzzy logic has made significant contributions in the construction industry due to its ability to handle uncertainties, imprecision, and subjective judgments that are inherent in construction decision-making processes. Below are some of its applications in the construction domain. Marandi et al. [38] presented a fuzzy-logic-based approach for assessing risks in construction projects. The authors proposed a methodology that incorporates fuzzy sets and fuzzy logic to handle the uncertainty and vagueness associated with risk assessment. This approach aims to provide decision-makers with a systematic and effective tool to evaluate and mitigate project risks. Andric et al. [39] focused on using fuzzy logic techniques to assess risks in road construction projects. The authors developed a fuzzy-logic-based model that considers multiple risk factors and their interrelationships. This model aims to provide a comprehensive risk assessment framework that can capture the imprecise and uncertain nature of road and belt construction project risks. Gajzler and Zima [40] developed a fuzzy-logic-based decision-making system for project risk management in construction. The authors proposed a model that incorporates fuzzy logic and expert knowledge to assess and rank project risks. This system aims to assist project managers in making informed decisions regarding risk mitigation strategies. Fayek [41] developed a fuzzy-logic-based decision-support system for construction project bidding. The authors proposed a model that uses fuzzy logic to evaluate bid proposals based on multiple criteria. This system aims to assist contractors in selecting the most suitable bid proposals by considering both qualitative and quantitative factors.

Govindan and Li [42] introduced a fuzzy-logic-based model for automating ergonomic risk assessment in construction projects. The authors proposed a methodology that combines fuzzy logic with a risk matrix to evaluate the likelihood and impact of risks. This model aims to provide a practical and intuitive approach to project risk assessment. Hendiani and Bagherpour [43] established an integrated indexed model for assessing social responsibility in the construction sector. The authors developed a comprehensive framework that integrates fuzzy logic to evaluate and prioritize social responsibility index. This model aims to provide a robust and flexible approach to project risk assessment. Obianyo et al. [44] presented a fuzzy decision-making framework for the cost assessment of overrun factors in the construction industry. This framework aims to assist decision-makers in identifying and addressing potential risks in construction projects.

2.6. Novelty and Research Gap

AI technologies have gained significant attention in the construction industry for their potential to improve efficiency, productivity, and decision-making. The combination of the Delphi method, ANP, and TOPSIS as a hybrid MCDM concept presents a novel approach to assess the role of AI in construction. Moreover, incorporating the fuzzy environment further enhances the applicability of this concept [45]. This section discusses the novelty of investigating the role of AI technologies in construction using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment, along with the research gap that it addresses.

2.6.1. Novelty

• Integration of methodologies: The combination of the Delphi method, ANP, and TOPSIS as a hybrid MCDM concept offers a novel approach to assess the role of AI technologies in the construction industry.
• Comprehensive evaluation: The hybrid concept allows for a holistic perspective on the impact and potential of AI technologies, as the Delphi method enables expert consensus-building, the ANP captures interdependencies among criteria, and TOPSIS facilitates the ranking of alternatives.
• Incorporation of a fuzzy environment: The inclusion of a fuzzy environment enhances the applicability of the concept by considering uncertainties, imprecise data, and subjective judgments that are commonly encountered in construction projects.

2.6.2. Research Gap

• Lack of integrated approach: Previous studies have focused on individual methodologies or different combinations, without considering the benefits of integrating the Delphi method, ANP, and TOPSIS. The research gap lies in exploring the role of AI technologies in construction using this hybrid MCDM concept.
• Handling uncertainty: Construction projects involve uncertainties, making it essential to consider a fuzzy environment. The research gap pertains to addressing uncertainty in decision-making processes related to AI technologies in construction.
• Comprehensive assessment: By adopting the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment, this research can provide a more realistic and robust evaluation of the role of AI technologies, filling the gap in the existing literature.

By integrating these methodologies and incorporating a fuzzy environment, this research addresses the research gap related to assessing the potential of AI technologies in construction. The findings can provide valuable insights for decision-makers and industry professionals, aiding in the adoption and implementation of AI in construction and improving overall industry performance. Further research in this area can advance the use of AI technologies in construction and enhance decision-making processes by leveraging the potential of AI in an informed and systematic manner.

2.7. Defining of Parameters and AI Alternatives

Parameters and alternatives are the core components of MCDM analysis that must be specified and analyzed in order to make informed judgments. Decision-makers can systematically assess options, examine numerous criteria, and make more accurate judgments based on their preferences and objectives by precisely specifying parameters and alternatives in MCDM analysis. This section emphasizes the significance of the specified parameters and AI alternatives for this ongoing MCDM investigation.

2.7.1. Parameters

Several factors can be taken into account when examining the impact of AI technologies in the construction sector [46]. These criteria offer a framework for evaluating and comprehending the potential and impact of AI in the construction industry. These criteria also represent different aspects or dimensions that are important for making a decision.

(a)

Technology criteria (TC): In the role of AI technologies in the construction industry, technology criteria refer to the specific requirements or considerations that need to be fulfilled for the effective implementation and utilization of AI in construction projects. These criteria help determine the complexity, aesthetics, value of data and algorithms, and advancements in innovations for the traditional problems in addressing the industry’s unique challenges and improving construction processes [21,22]. Below are some key technology sub-criteria relevant to AI in the construction industry.

• Complexity (CR1): The complexity sub-criterion, within the broader technology criteria, refers to the specific considerations related to the complexity of implementing and utilizing AI technologies in the construction industry. This sub-criterion helps assess the challenges and intricacies associated with incorporating AI solutions in construction processes. AI technologies can involve complex algorithms, machine learning models, and data processing techniques [5]. Assessing the technical complexity involves evaluating the levels of expertise, skills, and resources required to develop, deploy, and maintain AI solutions. This criterion considers factors such as the complexity of the AI architecture, algorithmic requirements, computational resources, and the availability of specialized technical talent. Integrating AI technologies with existing construction systems, software, and workflows can be complex. Construction companies need to assess the compatibility of AI solutions with their current infrastructure, including project management software, BIM platforms, and IoT devices [8]. This criterion examines the effort required to establish seamless integration, data-exchange protocols, and interoperability with other systems.
• Aesthetics (CR2): In the context of technology criteria for AI technologies in the construction industry, the aesthetics sub-criterion refers to the considerations related to the visual and user experience aspects of AI solutions. While aesthetics may not be the primary focus in construction, it plays a role in user acceptance, engagement, and the overall usability of AI technologies [11,12]. The user interface (UI) of AI applications should be visually appealing, intuitive, and user-friendly. A well-designed UI ensures that construction professionals can easily interact with the AI system, access information, and perform tasks efficiently. Visual elements, such as color schemes, typography, icons, and layout, should be carefully designed to provide a pleasant and engaging user experience.
• Value of data and algorithms (CR3): In the context of technology criteria for AI technologies in the construction industry, the value of data and algorithms sub-criterion refers to the considerations related to the quality, relevance, and effectiveness of the data and algorithms used in AI solutions. This sub-criterion plays a crucial role in determining the success and accuracy of AI applications. The value of data in AI applications depends on their quality, completeness, accuracy, and reliability [16]. Construction companies need to assess the availability of relevant and diverse datasets for training AI algorithms. High-quality data ensure that the AI system can learn effectively and produce reliable results.
• Advancements in innovations for the traditional problems (CR4): In the context of technology criteria for AI technologies in the construction industry, the sub-criterion of advancements in innovations for traditional problems refers to the consideration of how AI can contribute to addressing and solving longstanding challenges and issues in the construction industry [2,3]. This sub-criterion focuses on the transformative potential of AI in overcoming traditional problems and improving construction processes. One of the significant challenges in the construction industry is achieving higher efficiency and productivity. AI technologies can automate repetitive tasks, optimize resource allocation, and streamline workflows [18]. This sub-criterion assesses how AI innovations can improve project scheduling, optimize material management, automate documentation processes, and enhance overall construction efficiency.

(b)

Organization criteria (OC): These refer to the considerations relating to the organizational features and needs for the successful installation and utilization of AI solutions in the construction sector [30]. The organizational capabilities, preparation, and structure that are crucial for successfully implementing AI technologies in the construction process are the main focus of these sub-criteria. Below are some key organization sub-criteria relevant to AI in the construction industry.

• Government/management (CR5): The government/management sub-criterion within the organization criteria often concentrates on assessing how successfully AI technologies are implemented and managed at both the project and organizational levels in the context of the function of AI technologies in the construction sector. The organization’s capacity to create a strategic plan for incorporating AI technologies in building projects is evaluated by the strategic planning aspect. This involves defining goals, identifying relevant AI applications, and establishing a roadmap for implementation. Governance and policies examines the presence of governance structures and policies that guide the responsible and ethical use of AI technologies [12,13]. This includes privacy and data protection policies, compliance with regulations, and risk management strategies specific to AI implementation.
• Cost/sufficient budget (CR6): The cost/sufficient budget sub-factor within the organization criteria often focuses on assessing the financial aspects connected with the deployment of AI in the context of the role of AI technologies in the construction sector [47]. This sub-criterion examines the organization’s ability to allocate a sufficient budget for integrating AI technologies into construction projects. The budget planning aspect involves assessing the organization’s ability to accurately estimate the costs associated with AI’s implementation. This includes identifying the various cost components, such as hardware and software acquisition, data storage, infrastructure upgrades, training, and ongoing maintenance and support. Conducting a cost–benefit analysis helps evaluate the potential return on investment (ROI) of implementing AI technologies. This involves assessing the expected benefits, both tangible (e.g., increased productivity, improved quality, cost savings) and intangible (e.g., enhanced decision-making, reduced risk), and comparing them to the anticipated costs to determine the viability of implementing AI. Evaluating the organization’s procurement processes and vendor management strategies is important when considering the cost aspect [48]. This involves identifying suitable AI solution providers, conducting competitive bidding processes, negotiating contracts, and ensuring cost-effective vendor relationships to maximize the value for the allocated budget.
• Employee workforce (CR7): The employee workforce sub-criterion within the organization criteria often focuses on assessing the organization’s workforce in connection to the deployment of AI in the context of the usage of AI technologies in the construction sector. This sub-criterion looks at how the company manages its staff and ensures that they are prepared and adaptable to work with AI technologies [6,8]. The workforce planning aspect involves assessing the organization’s ability to plan for the integration of AI technologies by identifying the required workforce skills and competencies. This includes evaluating the organization’s understanding of how AI will impact different job roles and identifying the necessary workforce adjustments, such as upskilling initiatives. Evaluating the existing skills of the workforce is crucial for understanding the gaps that need to be addressed for the successful implementation of AI. This involves assessing the current skillsets of employees, identifying the areas where AI technologies will have the most impact, and determining the skills needed to effectively collaborate with and utilize AI tools and systems.
• Information exchange and communication/interoperability (CR8): The information exchange and communication/interoperability sub-criterion within the organization criteria typically focuses on assessing how effectively information is exchanged and communicated within the organization and with external stakeholders in the context of AI’s implementation. This is relevant to the role of AI technologies in the construction industry [49]. This sub-criterion examines the organization’s ability to ensure seamless data exchange and interoperability between the various systems and parties involved. Assessing the organization’s ability to integrate and standardize data from various sources is crucial for the effective implementation of AI. This involves evaluating whether the organization has processes in place to collect, aggregate, and preprocess data from different systems, sensors, and databases, ensuring the compatibility and quality of data for AI algorithms and models. Evaluating the organization’s ability to facilitate data sharing and collaboration among internal teams, project stakeholders, and partners is important [50]. This includes assessing whether the organization has established secure and efficient mechanisms for sharing data, insights, and project information, enabling effective collaboration and decision-making.
• Risk-taking ability (CR9): Risk-taking ability refers to an organization’s capacity and willingness to embrace and manage risks associated with adopting new technologies, such as AI, in the construction industry. This involves evaluating the organization’s attitude towards innovation, its adaptability to change, and its ability to handle potential risks and uncertainties. This aspect assesses the organization’s openness to new ideas and technologies. A construction company with strong risk-taking ability encourages innovative thinking and is willing to explore and experiment with AI solutions to improve its operations and processes [51]. The organization’s capability to manage and adapt to changes plays a vital role in determining its risk-taking ability. The adoption of AI in the construction industry often requires significant changes in workflows, skillsets, and organizational structures. A construction company with effective change-management strategies and a flexible approach is better equipped to embrace AI technologies and mitigate potential challenges. Implementing AI solutions involves investing in technology infrastructure, software, training, and human resources. Risk-taking ability considers the organization’s willingness to allocate sufficient resources to support the adoption and implementation of AI. This includes financial investments, skilled personnel, and time required for the integration process. The organization’s ability to identify and evaluate potential risks associated with the adoption of AI is crucial [52]. This involves conducting a comprehensive risk assessment, including technical, operational, legal, and ethical considerations. It also helps to develop risk mitigation strategies and contingency plans to address potential challenges and minimize negative impacts. A learning-oriented organizational culture promotes continuous improvement and encourages employees to acquire new skills and knowledge. Risk-taking ability encompasses fostering a culture of learning, where employees are encouraged to embrace new technologies, participate in training programs, and share knowledge and experiences related to the adoption of AI.

(c)

Environment criteria: The environment criteria often focus on assessing the environmental impact and sustainability factors connected with the implementation of AI in the context of its application in the construction industry [37]. These standards evaluate how AI innovations can support environmentally responsible behavior and sustainable construction methods. Below are some important factors related to the environment criteria.

• Upstream and downstream policy/laws (CR10): In the context of the role of AI technologies in the construction industry, the upstream and downstream policy/laws sub-criterion within the environment criteria focuses on evaluating the policies, regulations, and legal frameworks that govern the use of AI technologies and their impact on the environment in both the upstream (pre-construction) and downstream (post-construction) phases. Assessing the organization’s adherence to relevant policies and regulations regarding the use of AI technologies in the construction industry is important. This includes evaluating whether the organization complies with environmental laws, permits, and guidelines applicable to construction projects where AI technologies are employed [53]. Evaluating the organization’s compliance with regulations related to environmental impact assessments is crucial. This involves assessing whether the organization conducts thorough assessments of the potential environmental impacts of AI technologies throughout the construction lifecycle and incorporates mitigation measures as required by regulations.
• Trust between different companies/copyright/ownership (CR11): The trust between different companies, copyright, and ownership sub-criteria typically fall under the legal and ethical criteria in the context of the use of AI technologies in the construction sector. These sub-criteria concentrate on the legal and moral issues surrounding the application of AI and its effects on trust, copyright, and ownership rights in the building sector. This parameter helps to evaluate the establishment and maintenance of trust between different companies involved in AI-enabled construction projects [54,55]. It involves assessing whether companies have clear agreements, contracts, and partnerships that outline their roles, responsibilities, and obligations regarding AI technologies. Building trust ensures transparency, collaboration, and fair treatment among companies, fostering a conductive environment for the successful employment of AI.
• Social impacts (CR12): The social impacts sub-criterion within the social criteria focuses on evaluating the societal implications and effects of implementing AI technologies in the construction industry. It assesses how the adoption of AI influences various social aspects and stakeholders within and beyond the construction sector [56]. AI technologies in construction may impact the workforce by automating certain tasks and potentially displacing some job roles. Evaluating the organization’s efforts to mitigate workforce displacement, provide reskilling or upskilling opportunities, and ensure a smooth transition for workers affected by the implementation of AI is crucial. This involves considering the social implications of changing job requirements and the need for diverse skillsets. Assessing how AI technologies contribute to worker safety and health is important. AI can be utilized for risk assessment, safety monitoring, and providing real-time alerts to prevent accidents or hazards on construction sites. Evaluating the organization’s commitment to implementing AI systems that prioritize worker safety and health helps to create a safer work environment. Evaluating the organization’s efforts to engage and include stakeholders from diverse backgrounds is important. This involves considering whether the organization actively seeks input and feedback from workers, communities, and other relevant stakeholders impacted by the implementation of AI [57]. Promoting inclusivity and involving stakeholders in decision-making processes can lead to better outcomes and address potential social concerns.
• Regulatory measures (CR13): The regulatory measures sub-criterion focuses on evaluating the regulatory framework and measures in place to govern the use of AI technologies in the construction industry, particularly those pertaining to environmental considerations. It assesses whether organizations comply with the relevant regulations, permits, and guidelines related the implementation of AI and its environmental impact. Evaluating compliance with environmental impact assessment regulations is crucial [37,42]. It involves assessing whether organizations conduct thorough assessments of the potential environmental impacts of AI technologies in construction projects, and whether they implement mitigation measures as required by regulations. This ensures that the environmental implications of AI’s implementation are adequately evaluated and addressed. Assessing whether organizations comply with environmental laws and obtain the necessary permits and approvals is important [58]. This includes evaluating whether they meet the specific requirements related to the use of AI technologies in construction, such as emissions monitoring, waste management, and adherence to pollution control measures.

2.7.2. Alternatives

The construction industry has been using numerous AI technologies to boost productivity, safety, and process efficiency. In the context of the construction industry as used in this research, AI can be defined as the utilization of technologies to enhance numerous aspects of construction processes, operations, and decision-making. In the current study, AI mainly entails the employment of intelligent technologies and algorithms to automate processes, improve productivity, reduce errors, optimize resource allocation, detect faults, improve safety, and assist in informed decision-making throughout the building lifecycle. To serve these purposes, specialists identified three suitable prospective AI technologies from the prior studies listed below, which were examined for the current analysis.

(a)

Building information modeling (BIM) (A1): BIM is a digital representation of the physical and functional elements of a construction project, including 3D models, data, and collaboration tools. It enables stakeholders to visualize, analyze, and manage construction projects from design to operation. BIM facilitates efficient project management by improving collaboration, reducing errors, detecting clashes, scheduling, calculating costs and optimizing resource allocation.

(b)

Robotics and automation (A2): Robotics and automation technologies are used to automate repetitive and labor-intensive tasks in construction. They can perform repetitive tasks precisely and quickly, lowering labor requirements and increasing efficiency. This includes autonomous equipment; robotic arms for bricklaying, concrete pouring, and welding; drones for surveying and inspections; and automated machinery for material handling and assembly. Robotics enhances productivity, reduces risks, and improves construction precision.

(c)

Computer vision (A3): Cameras and image-processing techniques are used in computer vision to enable machines to comprehend visual inputs. Computer vision technology uses AI algorithms to analyze visual data, such as images and videos. Computer vision can be used on construction sites for quality control, safety monitoring, progress tracking, and object detection. It can identify potential hazards, detect defects or abnormalities, monitor worker compliance, and track project progress with greater accuracy.

3. Theoretical Framework

This section provides a step-by-step breakdown of the stated problem using three MCDM strategies to achieve the primary goal. It also emphasizes how the entire MCDM model is constructed from the start and contributes to the attainment of the goals [59,60]. However, before we go into the methodology, let us define the scope of the investigation. Every phase is thoroughly described in order to clarify the step-by-step flow of the entire decision-making process. Table 1 lists all of the parameters chosen for the ongoing analysis, and Figure 1 depicts the general design of the entire hybrid system.

Step 1: The first and most important step is to assemble a panel of experts who will make appropriate decisions and provide their perspectives on various aspects of the continuing MCDM investigation. A board of 30 members was constituted based on both the Indian and Chinese domains to reflect the insights of each country. The panel of 30 members was divided into three teams of 10 specialists each, as indicated in

Table 2. Background information of the panel experts.

No. of Expert Members	Professional Field	Designation	Experience	Country
Decision Team 1 (DT1)
2	Construction industry	Managing director	18	India
2	IT professional	Cloud engineer	13
1	Manufacturing industry	General manager	25
4	Construction industry	Chief engineer	28
1	Academic	Research project supervisor	22
Decision Team 2 (DT2)
4	Construction industry	Project manager	14	China
2	IT professional	Data analyst	13
1	IT professional	Technical lead	15
2	Construction industry	Senior construction manager	18
1	University	Professor	22
Decision Team 3 (DT3)
1	Transportation sector	Manager	16	India
2	Construction industry	Vice-president	17	India
2	Construction industry	Construction engineer	11	India
2	IT professional	Data scientist	19	China
2	Construction industry	Supervisor	12	China
1	IT professional	Senior programmer	15	China

Source: authors’ own elaboration.

. Out of the 30 members, 15 were selected from each country (India and China) in order to give equal importance to both countries. To ensure a transparent decision-making process, the first and second teams of 10 members each were completely based on individual countries (India and China, respectively), while the third team of 10 members was an amalgamation of experts from both regions, with 5 from India and 5 from China, as indicated in Table 2. By doing so, we increased the applicability of the present MCDM model to both countries. Region-specific information and viewpoints were gathered by involving local experts from each country, taking into account the distinctive characteristics and contextual variables of each region/country. Expert discussions can aid in refining the assessment of variable contributions and improving the accuracy of the analysis. Recognizing the heterogeneity in parameter contributions between countries, decision-makers may have tailored the MCDM framework accordingly. They may have created area- or country-specific models, or they may have adjusted the decision criteria to better represent the unique circumstances and priorities of each location. This adjustment enables a more precise portrayal of the variables in each context. In the case of the ANP, each team provided their own unique perspectives and comments on the pairwise comparison matrix [61]. The experts were only permitted to meet with members of their own team while performing pairwise judgements for the ANP method. There were no connections or connectivity between the three teams; thus, each team had to submit their judgments solely after speaking with their own team members. As a result, in the case of the ANP, there were three pairwise comparison matrices. The decisions of each group were fully hidden from the other groups. This decreased the risk of discrimination to some extent.

Step 2: The board members’ next task was to identify the most important parameters. Following an exhaustive brainstorming session, the professional members finalized all components of the cloud system. The decision-makers were highly experienced specialists with extensive knowledge and competence in the sector, and they selected the 13 important traits listed in Table 1 as having the greatest influence on AI infrastructure within the construction industry. Figure 2 depicts the breakdown of the 13 components into three fundamental groups. Figure 2 also depicts all of the variables and sub-factors taken into account for this investigation.

Step 3: The Delphi managerial technique was utilized to determine whether the selected 13 aspects were capable of achieving the expectations of the experts. Parameters with values greater than the acceptance degree level of the expert team were accepted for further investigation, while those with values less than the acceptance degree level were rejected.

Step 4: The parameters that met the Delphi screening test parameters were next entered into the ANP analysis stage, where the priority vectors of the final qualified parameters were evaluated. The ANP is especially good for determining criteria weights based on pairwise comparison, which can help the decision-maker to forecast the importance of each criterion [62]. The ANP is a generalized form of the AHP that constructs a hierarchical network considering all of the goals, conflicting parameters, and alternatives together into one decision-making framework.

Step 5: Finally, the ANP was followed by the TOPSIS MCDM technique to rate the three AI alternatives. TOPSIS was primarily used to find the preference ranking order of the alternatives, based on which the experts were able to make judgements about the best available AI option to be chosen for the construction industry.

3.1. Brainstorming Session with Expert Members

A council of 30 experts from various sectors was constituted and divided into three teams of 10 experts each. The members of the committee had extensive knowledge and expertise in their respective fields. The authors set one eligibility condition for assembling the committee: that all expert members must have at least 10 years of field experience. Personnel with more than 10 years of experience were only considered for making sound decisions based on their practical experience. As a result, each member of the board had extensive experience and competence in real life. Table 2 contains information on each committee member, such as their experience, profession, and designation. Following a thorough and in-depth analysis of numerous previously published articles in the field, several brainstorming stages were carried out to discover the parameters influencing AI technologies in the construction industry. Finally, during a Delphi brainstorming session with fellow team members, the professionals defined the 13 most significant criteria directly related to AI technologies. The first stage was to identify the majority of potential hurdles directly or indirectly resisting the implementation of AI technologies in the construction industry, as described in prior studies. Initially, the expert members compiled a list of 25 related AI factors that had been previously examined by many academics. In the second stage, the list was improved further by deleting extraneous weak components. It is also true that taking into consideration all of the parameters for computational analysis was not feasible, because the calculation would become complex and time-consuming. As a result, the committee members ultimately finalized 13 of the most relevant criteria from the list, for simplicity of mathematical computation. Table 1 depicts how the 13 traits were classified into three broad groups. Figure 2 displays the categorization of the 13 selected criteria into three major groups. The fundamental purpose of this research was to investigate the 13 components and three alternatives by applying the Delphi-ANP-TOPSIS hybrid MCDM model, where the Delphi method was used to measure the quality of the chosen factors, the ANP was used to evaluate the criteria weights, and TOPSIS ultimately helped in identifying the most appropriate option among the available choices and prescribing a preference ranking order of the alternatives.

3.2. Fuzzy Concept and Preliminaries

The fuzzy concept helps decision-makers to deal with and reflect uncertainty, ambiguity, and imprecision. Fuzzy logic is a mathematical framework that can be used to describe and record subjective judgments and linguistic expressions. In MCDM, decision-makers are frequently asked to express their preferences using qualitative or linguistic terms such as “high”, “medium”, or “low”. Fuzzy sets and fuzzy logic can be used by decision-makers to assign degrees of membership to these linguistic words, allowing for more expressive and flexible representation of subjective judgments. Imperfect or partial facts are common in real-world decision-making contexts [63]. Fuzzy logic can be used by decision-makers to model and handle erroneous data. Decision-makers can express the level of confidence or ambiguity associated with accessible information by assigning membership degrees to data points. This enables decision-makers to make informed choices even with the unavailability of reliable data. Decisions frequently have inherent ambiguity due to reasons such as a lack of information, changing conditions, or future events.

Decision-makers can more precisely estimate the degree of uncertainty by assigning membership degrees or possibility values to specific outcomes or situations. There are many different types of accessible fuzzy numbers, such as triangular, trapezoidal, Gaussian, and bell-shaped numbers, but triangular and trapezoidal fuzzy numbers (TFNs) are the most commonly used [64]. Figure 3 depicts TFN representations. The triangular form of fuzzy numbers was chosen for this continuing investigation due to its simplicity. When dealing with ambiguous or unclear content, triangular versions are easier to interpret and adjust to changes. They also provide straightforward mathematical operations and a clear grasp of the level of membership or possibilities. In MCDM, many criteria concepts are frequently evaluated. Decision-makers can utilize fuzzy logic to incorporate and resolve divergent criteria with overlapping or opposing preferences. Fuzzy aggregation methods, such as fuzzy weighted averages and fuzzy TOPSIS, enable decision-makers to aggregate fuzzy evaluations or rankings based on several criteria in a logical and consistent way [65]. By adjusting the level of fuzziness or ambiguity in the input data or parameters, decision-makers can evaluate the sensitivity of the consequences and the robustness of the choice under varied conditions.

This analysis supports decision-makers in understanding the stability and reliability of the decision-making process. Fuzzy concepts are useful in MCDM because they allow decision-makers to cope with subjective judgments, imprecise data, ambiguity, and a range of criteria in a more expressive and flexible manner [66]. The membership function denoted by ${\mu }_{\tilde{\mathrm{F}}}$(x) in Figure 3 is a mathematical function that confers a degree of membership to each value within the universe of discourse, based on its interaction with the fuzzy number. This determines the degree and magnitude of the fuzzy number’s given value. The membership function for a fuzzy number assigns membership values ranging from 0 to 1 to entities in the discourse universe. It quantifies the extent to which each value in the discourse universe belongs to the fuzzy number [67]. Fuzzy number operations such as addition, subtraction, multiplication, and division can be performed by converting the membership function to quantitative values and analyzing uncertainty.

Figure 3 depicts the bottom bound “e”, middle bound “f”, and upper bound “g” of the triangular fuzzy number $\tilde{\mathrm{F}}$ = (e, f, g). Similarly, for trapezoidal fuzzy number $\tilde{\mathrm{F}}$ = (e, f1, f2, g), the values “e” and “g” represent the bottom and upper bounds, while “f1” and “f2” represent the intermediate bounds. Equations (1) and (2) can be used to represent the membership functions of triangular and trapezoidal fuzzy numbers, respectively:

$${\mu }_{\tilde{\mathrm{F}}}(x) \left\{\begin{array}{c}0 \mathrm{for} x<e\\ \frac{x-\mathrm{e}}{\mathrm{f}-\mathrm{e}} \mathrm{for} e\le x\le f\\ \frac{\mathrm{g}-x}{\mathrm{g}-\mathrm{f}} \mathrm{for} f\le x\le g\\ 0 \mathrm{for} \mathrm{otherwise}\end{array}\right.$$

(1)

$${\mu }_{\tilde{\mathrm{F}}}(x) \left\{\begin{array}{c}0 \mathrm{for} x<e\\ \frac{x-\mathrm{e}}{{\mathrm{f}}_1-\mathrm{e}} \mathrm{for} e\le x\le {\mathrm{f}}_1\\ 1 \mathrm{for} {\mathrm{f}}_1\le x\le {\mathrm{f}}_2\\ \frac{\mathrm{g}-x}{\mathrm{g}-{\mathrm{f}}_2} \mathrm{for} {\mathrm{f}}_2\le x\le g\\ 0 \mathrm{for} \mathrm{otherwise}\end{array}\right.$$

(2)

Now consider some of the mathematical operations of two fuzzy integers $\widetilde{{\mathrm{F}}_1}$ = (e₁, f₁, g₁) and $\widetilde{{\mathrm{F}}_2}$ = (e₂, f₂, g₂):

$$\mathrm{Addition} \to \left\{\begin{array}{c}\widetilde{{\mathrm{F}}_1}+\tilde{{\tilde{\mathrm{F}}}_2}=\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)+\left({\mathrm{e}}_2, {\mathrm{f}}_2, {\mathrm{g}}_2\right)=\left({\mathrm{e}}_1+{\mathrm{e}}_2, { \mathrm{f}}_1+{\mathrm{f}}_2, { \mathrm{g}}_1+{\mathrm{g}}_2\right)\\ \widetilde{{\mathrm{F}}_1}+\lambda =\left({\mathrm{e}}_1+\lambda , { \mathrm{f}}_1+\lambda , { \mathrm{g}}_1+\lambda \right)\end{array}\right.$$

(3)

$$\mathrm{Subtraction} \to \left\{\begin{array}{c}\widetilde{{\mathrm{F}}_1}-\widetilde{{\mathrm{F}}_2}=\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)-\left({\mathrm{e}}_2, {\mathrm{f}}_2, {\mathrm{g}}_2\right)=\left({\mathrm{e}}_1-{\mathrm{g}}_2, { \mathrm{f}}_1-{\mathrm{f}}_2, { \mathrm{g}}_1-{\mathrm{e}}_2\right)\\ \widetilde{{\mathrm{F}}_1}-\lambda =\left({\mathrm{e}}_1-\lambda , { \mathrm{f}}_1-\lambda , { \mathrm{g}}_1-\lambda \right)\end{array}\right.$$

(4)

$$\mathrm{Multiplication} \to \left\{\begin{array}{c}\widetilde{{\mathrm{F}}_1}\times \widetilde{{\mathrm{F}}_2}=\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)\times \left({\mathrm{e}}_2, {\mathrm{f}}_2, {\mathrm{g}}_2\right)=\left({\mathrm{e}}_1\times {\mathrm{e}}_2, { \mathrm{f}}_1\times {\mathrm{f}}_2, { \mathrm{g}}_1\times {\mathrm{g}}_2\right)\\ \widetilde{{\mathrm{F}}_1}\times \lambda =\left({\mathrm{e}}_1\times \lambda , { \mathrm{f}}_1\times \lambda , { \mathrm{g}}_1\times \lambda \right)\end{array}\right.$$

(5)

$$\mathrm{Division} \to \left\{\begin{array}{c}\widetilde{{\mathrm{F}}_1}÷\widetilde{{\mathrm{F}}_2}=\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)÷\left({\mathrm{e}}_2, {\mathrm{f}}_2, {\mathrm{g}}_2\right)=\left(\frac{{\mathrm{e}}_1}{{\mathrm{g}}_2}, \frac{{\mathrm{f}}_1}{{\mathrm{f}}_2}, \frac{{\mathrm{g}}_1}{{\mathrm{e}}_2}\right)\\ \widetilde{{\mathrm{F}}_1}÷\lambda =\left(\frac{{\mathrm{e}}_1}{\mathsf{\lambda }}, \frac{{\mathrm{f}}_1}{\mathsf{\lambda }}, \frac{{\mathrm{g}}_1}{\mathsf{\lambda }}\right)\end{array}\right.$$

(6)

$$\mathrm{Inverse} \to \left\{\frac{1}{\widetilde{{\mathrm{F}}_1}}\right.=\frac{1}{\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)}=(\frac{1}{{\mathrm{g}}_1}, \frac{1}{{\mathrm{f}}_1}, \frac{1}{{\mathrm{e}}_1})$$

(7)

• The fuzzy geometric mean value of “k” fuzzy numbers {${\tilde{\mathrm{F}}}_1$ = $\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)$, ${\tilde{\mathrm{F}}}_2$ = $\left({\mathrm{e}}_2,{ \mathrm{f}}_2,{ \mathrm{g}}_2\right)$, …… ${\tilde{\mathrm{F}}}_{\mathrm{k}}$ = $\left({\mathrm{e}}_{\mathrm{k}},{ \mathrm{f}}_{\mathrm{k}},{ \mathrm{g}}_{\mathrm{k}}\right)$} can be computed as follows:

  $$\widetilde{\mathrm{GM}} (\widetilde{{\mathrm{F}}_1}, \widetilde{{\mathrm{F}}_2}\dots ., \widetilde{{\mathrm{F}}_{\mathrm{k}}})=\left({\left({{\displaystyle \prod }}_{i=1}^k{e}_i\right)}^{\frac{1}{k}}, {\left({{\displaystyle \prod }}_{i=1}^k{f}_i\right)}^{\frac{1}{k}}, {\left({{\displaystyle \prod }}_{i=1}^k{g}_i\right)}^{\frac{1}{k}}\right)$$

  (8)
• Defuzzification of a fuzzy number ${\tilde{\mathrm{F}}}_1$ = $\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)$:

  $$\mathrm{Defuzzify} ({\mathrm{F}}_1)=\frac{{\mathrm{e}}_1+{\mathrm{f}}_1+{\mathrm{g}}_1}{3}$$

  (9)
• If there are “k” decision-makers expressing their views, then the fuzzy numbers {${\tilde{\mathrm{F}}}_1$ = $\left({\mathrm{e}}_1,{ \mathrm{f}}_1,{ \mathrm{g}}_1\right)$, ${\tilde{\mathrm{F}}}_2$ = $\left({\mathrm{e}}_2,{ \mathrm{f}}_2,{ \mathrm{g}}_2\right)$, …… ${\tilde{\mathrm{F}}}_{\mathrm{k}}$ = $\left({\mathrm{e}}_{\mathrm{k}},{ \mathrm{f}}_{\mathrm{k}},{ \mathrm{g}}_{\mathrm{k}}\right)$} can be aggregated as follows:

  $$\tilde{\mathrm{A}} (\widetilde{{\mathrm{F}}_1}, \widetilde{{\mathrm{F}}_2}\dots ., \widetilde{{\mathrm{F}}_{\mathrm{k}}})=\left(\begin{array}{c}\min {\mathrm{e}}_{\mathrm{i}}\\ \mathrm{k}\end{array}, \frac{1}{\mathrm{k}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{k}}{\mathrm{f}}_{\mathrm{i}}, \begin{array}{c}\max {\mathrm{g}}_{\mathrm{i}}\\ \mathrm{k}\end{array}\right)$$

  (10)

4. Materials and Methods

This section focuses on the computational analysis and mathematical computations of the MCDM models utilized in this study, which include the Delphi method, the ANP, and TOPSIS. The basic outline of the entire study is already presented in the prior sections, explaining all of the essential elements that influence AI technologies in the construction industry. All of these tools will now be used to assess the performance of each of the elements and to rate the alternatives. The Delphi method is used initially, followed by the ANP and TOPSIS, to examine the suitability of the 13 selected parameters [68]. The ANP is used to compute the criteria weights, and TOPSIS is used to propose the preference ranking order of the three available AI technologies. The computing specifics for all of the applied MCDM investigations are given in the subsections that follow. Let us begin with the Delphi analysis and slowly move into ANP and TOPSIS one by one.

4.1. Data Analysis of the Delphi Method

Delphi analysis is a structured and iterative forecasting or decision-making approach that enlists the help of a panel of experts to provide insights and predictions. When dealing with complex problems that lack complete information, it is widely used [68,69]. The Delphi analysis can be performed using the steps outlined below.

Step 1: The first step is to form an expert panel, as illustrated in Table 2. A group of 30 experts, typically individuals with competence and aptitude in the relevant topic or problem area, are identified and invited to participate in the Delphi analysis. To conduct the Delphi session, the 30 panel members were divided into 10 small groups of three decision-makers each and asked to express their thoughts on the linguistic relevance of each of the 13 criteria mentioned in

Table 3. Qualitative judgements of the experts for Delphi analysis.

Criteria	India					China
	Round 1	Round 2	Round 3	Round 4	Round 5	Round 6	Round 7	Round 8	Round 9	Round 10
CR1	VHI	HI	HI	MI	VHI	VHI	EHI	HI	VHI	EHI
CR2	ELI	ELI	MI	VLI	ELI	ELI	LI	MI	MI	ELI
CR3	VHI	EHI	VHI	HI	MI	HI	HI	HI	VHI	MI
CR4	VHI	MI	EHI	HI	VHI	VHI	EHI	HI	EHI	MI
CR5	HI	HI	VHI	MI	HI	MI	HI	MI	HI	VHI
CR6	HI	HI	EHI	HI	HI	HI	VHI	HI	MI	EHI
CR7	LI	MI	ELI	LI	ELI	MI	MI	VLI	VLI	ELI
CR8	VHI	VHI	EHI	MI	EHI	MI	MI	MI	HI	HI
CR9	HI	HI	VHI	MI	MI	VHI	MI	HI	MI	EHI
CR10	MI	HI	MI	HI	HI	EHI	MI	VHI	HI	EHI
CR11	HI	HI	VHI	MI	MI	VHI	HI	VHI	EHI	MI
CR12	EHI	EHI	VHI	VHI	MI	HI	EHI	MI	MI	VHI
CR13	MI	MI	VLI	ELI	LI	ELI	LI	ELI	ELI	VLI

Source: interpreted by the expert team members.

. It should be emphasized that 5 of the 10 were from the Indian context, while the other 5 were of Chinese origin. By combining expert opinions from both locations, the analysis was made universal and applicable to both countries.

Step 2: The experts are given a questionnaire or survey in the first round. The questionnaire consists of open-ended questions or statements on the current research topic. Simultaneously, the 10 expert groups are requested to provide their independent judgements, predictions, or ideas in the form of qualitative expressions in 10 rounds. The relevance of the chosen parameters is conveyed by the experts’ voiced comments. Table 3 displays the qualitative judgments made by the 10 decision-making groups.

Step 3: The experts’ verbal expressions are quantified into numerical values using the scale shown in

Table 4. Quantification of verbal interpretation into numeric values and TFNs.

For Delphi Analysis				For ANP Analysis
Qualitative Terms	Notations	Crisp Value	Fuzzy Values	Qualitative Terms	Notations	Crisp Value	Fuzzy Values
Extremely low importance	ELI	1	1, 1, 1	Equal importance	EI	1	1, 1, 1
Very low importance	VLI	2	1, 2, 3	Very low importance	VLI	2	1, 2, 3
Low importance	LI	3	2, 3, 4	Low importance	LI	3	2, 3, 4
Moderate importance	MI	5	4, 5, 6	Moderate importance	MI	5	4, 5, 6
High importance	HI	7	6, 7, 8	High importance	HI	7	6, 7, 8
Very high importance	VHI	8	7, 8, 9	Very high importance	VHI	8	7, 8, 9
Extremely high importance	EHI	9	9, 9, 9	Extremely high importance	EHI	9	9, 9, 9
Fuzzy geometric mean value (FGMV)				3.142, 3.954, 4.645
Defuzzification (acceptance degree level for Delphi analysis)				3.914

Source: authors’ own elaboration.

, which are then translated into their respective triangular fuzzy numbers. Figure 4 depicts the transformations of crisp numeric values into TFNs, whereas

Table 5. Final results of Delphi analysis.

MC Indicators	SC Indicators	Round 1	Round 2	Round 3	Round 4	Round 5	Round 6	Round 7	Round 8	Round 9	Round 10	FGMV	AIN	Status	Final Indicators
TC	CR1	7, 8, 9	6, 7, 8	6, 7, 8	4, 5, 6	7, 8, 9	7, 8, 9	9, 9, 9	6, 7, 8	7, 8, 9	9, 9, 9	6.646, 7.508, 8.342	7.499	Accept	TC1
	CR2	1, 1, 1	1, 1, 1	4, 5, 6	1, 2, 3	1, 1, 1	1, 1, 1	2, 3, 4	4, 5, 6	4, 5, 6	1, 1, 1	1.625, 1.939, 2.195	1.919	Reject	-
	CR3	7, 8, 9	9, 9, 9	7, 8, 9	6, 7, 8	4, 5, 6	6, 7, 8	6, 7, 8	6, 7, 8	7, 8, 9	4, 5, 6	6.034, 6.985, 7.917	6.979	Accept	TC2
	CR4	7, 8, 9	4, 5, 6	9, 9, 9	6, 7, 8	7, 8, 9	7, 8, 9	9, 9, 9	6, 7, 8	9, 9, 9	4, 5, 6	6.544, 7.345, 8.106	7.332	Accept	TC3
OC	CR5	6, 7, 8	6, 7, 8	7, 8, 9	4, 5, 6	6, 7, 8	4, 5, 6	6, 7, 8	4, 5, 6	6, 7, 8	7, 8, 9	5.479, 6.499, 7.513	6.497	Accept	OC1
	CR6	6, 7, 8	6, 7, 8	9, 9, 9	6, 7, 8	6, 7, 8	6, 7, 8	7, 8, 9	6, 7, 8	4, 5, 6	9, 9, 9	6.345, 7.213, 8.053	7.204	Accept	OC2
	CR7	2, 3, 4	4, 5, 6	1, 1, 1	2, 3, 4	1, 1, 1	4, 5, 6	4, 5, 6	1, 2, 3	1, 2, 3	1, 1, 1	1.741, 2.319, 2.814	2.291	Reject	-
	CR8	7, 8, 9	7, 8, 9	9, 9, 9	4, 5, 6	9, 9, 9	4, 5, 6	4, 5, 6	4, 5, 6	6, 7, 8	6, 7, 8	5.706, 6.608, 7.474	6.596	Accept	OC3
	CR9	6, 7, 8	6, 7, 8	7, 8, 9	4, 5, 6	4, 5, 6	7, 8, 9	4, 5, 6	6, 7, 8	4, 5, 6	9, 9, 9	5.479, 6.444, 7.387	6.437	Accept	OC4
EC	CR10	4, 5, 6	6, 7, 8	4, 5, 6	6, 7, 8	6, 7, 8	9, 9, 9	4, 5, 6	7, 8, 9	6, 7, 8	9, 9, 9	5.851, 6.744, 7.602	6.732	Accept	EC1
	CR11	6, 7, 8	6, 7, 8	7, 8, 9	4, 5, 6	4, 5, 6	7, 8, 9	6, 7, 8	7, 8, 9	9, 9, 9	4, 5, 6	5.795, 6.754, 7.693	6.747	Accept	EC2
	CR12	9, 9, 9	9, 9, 9	7, 8, 9	7, 8, 9	4, 5, 6	6, 7, 8	9, 9, 9	4, 5, 6	4, 5, 6	7, 8, 9	6.284, 7.102, 7.876	7.087	Accept	EC3
	CR13	4, 5, 6	4, 5, 6	1, 2, 3	1, 1, 1	2, 3, 4	1, 1, 1	2, 3, 4	1, 1, 1	1, 1, 1	1, 2, 3	1.516, 1.974, 2.352	1.947	Reject	-

Source: authors’ own elaboration.

depicts the fuzzy conversions of qualitative ratings.

Step 4: The analysis and feedback from the initial set of responses are collected, summarized, and anonymized at this time. The experts are then given feedback in the form of a report highlighting areas of agreement, disagreement, or divergence in their viewpoints. This encourages professionals to rethink their initial conclusions in light of the broader consensus.

Step 5: The experts are invited to participate in subsequent rounds based on the input they have received. The experts are given the opportunity of amending or modifying their initial comments in light of other experts’ thoughts and viewpoints. This process is repeated until there is agreement or convergence of viewpoints, or until a predefined stopping point is achieved. In this case, 10 rounds of interviews were conducted to eliminate the possibility of prejudice. The 30 participants were divided into 10 small groups of three specialists each, as previously mentioned. In this way, each group can be assumed as a separate round, and each group’s three experts delivered their conclusions while remaining entirely anonymous to the other group members. However, the final judgments of the 10 rounds are reported in Table 3.

Step 6: As stated in Table 5, Equation (8) is now utilized to calculate the fuzzy geometric mean value (FGMV) by aggregating all 10 round judgements.

Step 7: Defuzzification is now executed, and the acceptance index (AIN) of each parameter in Table 5 is computed using Equation (9).

Step 8: Based on the geometric mean, the committee board members decided the acceptance degree level to be 3.914, as shown in Table 4. The experts could now select parameters from Table 5 with values greater than 3.914 and reject those with values less than this acceptance degree level. The parameters that met the Delphi screening technique are indicated in Table 5. These factors were finalized for the subsequent ANP and TOPSIS analyses. Table 5 clearly shows that aesthetics (CR2), employee workforce (CR7), and regulatory measures (CR13) failed to meet the expectations and were thus rejected, whereas 10 parameters qualified for the next stage of the analysis and were regarded as the final parameters that mostly influence the AI technologies within the construction industry. As a result, four SCs under OC and three SCs each under TC and EC finally qualified for the second round, as indicated in Table 5.

Step 9: As the Delphi technique concludes, the final round’s findings are analyzed to identify common themes, patterns, or areas of agreement among the experts. To present the findings, summary statistics, distributions, or consensus statements are sometimes employed.

Delphi sessions’ systematic and iterative character allows for the emergence and convergence of expert viewpoints over numerous rounds. This fosters collective intelligence and offers a systematic approach to dealing with ambiguity and complexity in decision-making or forecasting. Delphi analysis is widely utilized in many domains, such as technical forecasting, policy creation, strategic planning, risk assessment, and project management. It is especially useful when dealing with complex issues that necessitate input from various sources, and when standard techniques are insufficient or impractical.

4.2. Data Analysis of the ANP

The ANP is an extension of the AHP developed by Thomas L. Saaty. While the AHP focuses on hierarchies with a single level of criteria and alternatives, the ANP allows for more complex decision problems that involve interdependencies and feedback among elements in the hierarchy. The ANP extends the AHP by introducing the concept of a network, where elements in the hierarchy can be connected in a network-like structure. This allows for a more comprehensive analysis of complex decision situations that involve interactions and dependencies among criteria, sub-criteria, and alternatives [56]. The ANP process involves the following steps for its operation:

Step 1 (define the decision problem): Clearly articulate the decision problem and identify the criteria, sub-criteria, and alternatives involved, as portrayed in Figure 5.

Step 2 (establish a network): Construct a network structure that represents the decision problem, incorporating both the hierarchy and the interdependencies among elements, as clearly depicted in Figure 5. The network can have direct and indirect connections between elements, forming a network of influences.

Step 3 (pairwise comparisons): According to Equation (11), create a pairwise comparison matrix (n_i $\times$ n_j), similar to the AHP. Pairwise comparison matrices are used to assess the relative importance or preference of elements. The parameter “n” represents the total number of parameters investigated for the pairwise comparison, and the parameter “p_ij” reflects the decision components of the pairwise matrix displayed in

Table 6. Pairwise judgements for MC and SC.

		Decision Team	MC			SC within TC			SC within OC				SC within EC
			TC	OC	EC	TC1	TC2	TC3	OC1	OC2	OC3	OC4	EC1	EC2	EC3
MC	TC	DT1	EI	HI	VLI
		DT2	EI	VHI	LI
		DT3	EI	HI	MI
	OC	DT1		EI
		DT2		EI
		DT3		EI
	EC	DT1		MI	EI
		DT2		LI	EI
		DT3		LI	EI
SC within TC	TC1	DT1				EI
		DT2				EI
		DT3				EI
	TC2	DT1				HI	EI	LI
		DT2				MI	EI	LI
		DT3				MI	EI	LI
	TC3	DT1				VLI		EI
		DT2				VLI		EI
		DT3				LI		EI
SC within OC	OC1	DT1							EI
		DT2							EI
		DT3							EI
	OC2	DT1							EHI	EI	MI	VHI
		DT2							EHI	EI	HI	HI
		DT3							EHI	EI	MI	HI
	OC3	DT1							MI		EI	LI
		DT2							LI		EI	LI
		DT3							LI		EI	VLI
	OC4	DT1							VLI			EI
		DT2							VLI			EI
		DT3							VLI			EI
SC within EC	EC1	DT1											EI
		DT2											EI
		DT3											EI
	EC2	DT1											VHI	EI	VLI
		DT2											HI	EI	LI
		DT3											VHI	EI	VLI
	EC3	DT1											VLI		EI
		DT2											VLI		EI
		DT3											LI		EI

Source: interpreted by the expert team members.

. To achieve the local and global weights, the three main criteria TC, OC, and EC are compared with one another, followed by the sub-criteria comparisons within each MC, as shown in Table 6. Hence, each pairwise judgement is accompanied by three decision teams in qualitative terms, as clearly presented in Table 6. These qualitative phrases are transformed into quantitative values using the scale provided in Table 4, followed by their respective TFNs. Similarly, the three alternatives considered for this analysis are also compared with one another with respect to each SC shown in

Table 7. Pairwise judgements of three alternatives with respect to each SC.

OC
OC		Decision team	OC1			OC2			OC3			OC4
			A1	A2	A3	A1	A2	A3	A1	A2	A3	A1	A2	A3
	A1	DT1	EI	LI		EI	LI		EI	VLI		EI	VLI
		DT2	EI	LI		EI	VLI		EI	LI		EI	LI
		DT3	EI	VLI		EI	VLI		EI	VLI		EI	VLI
	A2	DT1		EI			EI			EI			EI
		DT2		EI			EI			EI			EI
		DT3		EI			EI			EI			EI
	A3	DT1	MI	VHI	EI	MI	EHI	EI	LI	EHI	EI	VLI	VHI	EI
		DT2	MI	HI	EI	MI	VHI	EI	MI	VHI	EI	LI	HI	EI
		DT3	MI	HI	EI	LI	VHI	EI	LI	EHI	EI	LI	HI	EI
Consistency ratio (CR)			0.028			0.004			0.002			0.003
TC												RI values
TC		Decision team	TC1			TC2			TC3			n		RI
	A1	DT1	EI			EI			EI			1		0
		DT2	EI			EI			EI			2		0
		DT3	EI			EI			EI			3		0.58
	A2	DT1	MI	EI	LI	EHI	EI	MI	VHI	EI	MI	4		0.90
		DT2	MI	EI	VLI	VHI	EI	LI	VHI	EI	MI	5		1.12
		DT3	MI	EI	LI	EHI	EI	MI	EHI	EI	MI	6		1.24
	A3	DT1	VLI		EI	VLI		EI	LI		EI	7		1.32
		DT2	VLI		EI	VLI		EI	VLI		EI	8		1.41
		DT3	LI		EI	LI		EI	VLI		EI	9		1.45
Consistency ratio (CR)			0.005			0.004			0.016			10		1.49
EC												11		1.51
EC		Decision team	EC1			EC2			EC3			12		1.54
	A1	DT1	EI	LI	MI	EI	MI	HI	EI	LI	HI	13		1.56
		DT2	EI	LI	LI	EI	MI	MI	EI	LI	HI	14		1.57
		DT3	EI	VLI	MI	EI	LI	HI	EI	LI	HI	15		1.59
	A2	DT1		EI	VLI		EI	VLI		EI	LI	16		1.6
		DT2		EI	VLI		EI	VLI		EI	LI	17		1.61
		DT3		EI	VLI		EI	VLI		EI	LI	18		1.61
	A3	DT1			EI			EI			EI	19		1.62
		DT2			EI			EI			EI	20		1.63
		DT3			EI			EI			EI	21		1.63
Consistency ratio (CR)			0.005			0.008			0.006			22		1.64

Source: interpreted by the expert team members.

. Hence, for 10 SCs, there will be 10 pairwise comparisons among the three alternatives, and for each comparison there will be three judgements from the three decision teams. Table 7 clearly represents the alternatives’ pairwise judgements for each of the 10 cases.

$$\mathrm{P} ({\mathrm{n}}_{\mathrm{i}}\times {\mathrm{n}}_{\mathrm{j}})=\left[\begin{array}{c}{\mathrm{p}}_{11}\\ {\mathrm{p}}_{21}\\ \dots \\ {\mathrm{p}}_{\mathrm{i}1}\end{array} \begin{array}{c}{\mathrm{p}}_{12}\\ {\mathrm{p}}_{22}\\ \dots \\ {\mathrm{p}}_{\mathrm{i}2}\end{array} \begin{array}{c}\dots \\ \dots \\ \dots \\ \dots \end{array} \begin{array}{c}{\mathrm{p}}_{1\mathrm{j}}\\ {\mathrm{p}}_{2\mathrm{j}}\\ \dots \\ {\mathrm{p}}_{\mathrm{ij}}\end{array}\right]$$

(11)

The three pairwise matrices are aggregated using Equation (10) in each case to create the fuzzified pairwise comparison matrix, which is then defuzzified using Equation (9), providing the final pairwise matrix.

Step 4 (normalization): Normalization is performed using Equation (12) to stabilize the given data for pairwise comparisons. ${‘\mathrm{N}}_{\mathrm{ij}}^{\mathrm{AHP}}’$ represents the normalized values.

$${\mathrm{N}}_{\mathrm{ij}}^{\mathrm{ANP}}=\frac{{\mathrm{p}}_{\mathrm{ij}}}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{p}}_{\mathrm{ij}}}$$

(12)

Step 5 (derive priority weights): Based on the pairwise comparisons, calculate the priority weights (w_j) for each element in the network using Equation (13). These weights reflect the relative importance of the elements, considering both direct and indirect influences.

Table 8. Summary of criteria weights from the ANP analysis.

MC	MC Weights	MC CR	MC RANK	SC	SC Local Weights	SC CR	SC Local Rank	SC Global Weights	SC Global Rank
Technology criteria (TC)	0.679	0.033	1	Complexity of AI (TCI)	0.104	0.005	3	0.071	4
				Value of data and algorithms (TC2)	0.659		1	0.448	1
				Advancements in innovations for the traditional problems (TC3)	0.237		2	0.161	2
Organization criteria (OC)	0.077		3	Government/management (OC1)	0.055	0.040	4	0.004	10
				Cost/sufficient budget (OC2)	0.675		1	0.052	6
				Information exchange and communication/interoperability (OC3)	0.180		2	0.014	8
				Risk-taking ability (OC4)	0.090		3	0.007	9
Environment criteria (EC)	0.244		2	Upstream and downstream policy/laws (EC1)	0.094	0.005	3	0.023	7
				Trust between different companies/copyright/ownership (EC2)	0.657		1	0.160	3
				Social impacts (EC3)	0.249		2	0.061	5

Source: authors’ own elaboration.

depicts the factor weights of MC and SC, evaluated on the basis of Table 6. Similarly, priority vectors of the three alternatives in each case are also calculated on the basis of Table 7 and illustrated in

Table 9. Supermatrix for the ANP.

	Goal	TC1	TC2	TC3	OC1	OC2	OC3	OC4	EC1	EC2	EC3	A1	A2	A3
Goal	0	0	0	0	0	0	0	0	0	0	0	0	0	0
TC1	0.071	0	0	0	0	0	0	0	0	0	0	0	0	0
TC2	0.448	0	0	0	0	0	0	0	0	0	0	0	0	0
TC3	0.161	0	0	0	0	0	0	0	0	0	0	0	0	0
OC1	0.004	0	0	0	0	0	0	0	0	0	0	0	0	0
OC2	0.052	0	0	0	0	0	0	0	0	0	0	0	0	0
OC3	0.014	0	0	0	0	0	0	0	0	0	0	0	0	0
OC4	0.007	0	0	0	0	0	0	0	0	0	0	0	0	0
EC1	0.023	0	0	0	0	0	0	0	0	0	0	0	0	0
EC2	0.160	0	0	0	0	0	0	0	0	0	0	0	0	0
EC3	0.061	0	0	0	0	0	0	0	0	0	0	0	0	0
A1	0	0.116	0.083	0.081	0.180	0.190	0.195	0.244	0.607	0.705	0.669	1	0	0
A2	0	0.622	0.727	0.746	0.084	0.084	0.084	0.094	0.256	0.190	0.243	0	1	0
A3	0	0.263	0.190	0.173	0.736	0.726	0.721	0.661	0.138	0.105	0.088	0	0	1

Source: authors’ own elaboration.

.

$${\mathrm{w}}_{\mathrm{j}}=\frac{{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{N}}_{\mathrm{ij}}^{\mathrm{ANP}}}{\mathrm{n}}$$

(13)

Step 6 (consistency check): Assess the consistency of the pairwise comparisons to ensure logical and coherent judgments. Inconsistencies can be resolved through adjustments and iterations. All of the pairwise comparisons presented in Table 6 and Table 7 are then validated by computing the CR values. This approach is divided into two stages, computing the CI and determining the CR. The CI and CR can be calculated using Equation (14) and Equation (15), respectively. The CR value determines whether the pairwise judgements are consistent or not. The maximum permissible CR limit is 0.1 or 10%. If the CR value is less than 0.1 or 10%, the qualitative judgements of the three expert teams proposed in Table 6 and Table 7 may be considered to be consistent. Similarly, CR > 0.1 indicates an inconsistent pairwise matrix; in this situation, the expert opinions must be amended until the CR value falls within the acceptable range of 0.1. For Table 6, the CR values of all of the comparisons are evaluated in Table 8; similarly the CR values of all of the comparisons for Table 7 are shown in Table 7 itself. It is evident that all of the expert’s judgements were found to be consistent, and that the CR values for every scenario lie well within the permissible limit. As a result, the priority weights evaluated in Step 5 can be treated as the final criteria weights.

$$\mathrm{CI}=\frac{\left({\mathsf{\lambda }}_{\max }- \mathrm{n}\right)}{\left(\mathrm{n}-1\right)}$$

(14)

$$\mathrm{CR}=\frac{\mathrm{CI}}{\mathrm{RI}}$$

(15)

The notation “RI” stands for a randomly generated index, the values of which can be adopted from Table 7. Table 8 summarizes all of the criteria related to the final outcomes, including local weights, global weights, local rank, global rank, and CR values.

Step 7 (supermatrix calculation): Combine the priority weights and interdependency relationships obtained from the previous stages to form a supermatrix, as shown in Table 9, which represents the overall structure of the decision problem.

Step 8 (eigenvector calculation): Calculate the eigenvector matrix shown in

Table 10. Limit matrix.

	Goal	TC1	TC2	TC3	OC1	OC2	OC3	OC4	EC1	EC2	EC3	A1	A2	A3
Goal	0	0	0	0	0	0	0	0	0	0	0	0	0	0
TC1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
TC2	0	0	0	0	0	0	0	0	0	0	0	0	0	0
TC3	0	0	0	0	0	0	0	0	0	0	0	0	0	0
OC1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
OC2	0	0	0	0	0	0	0	0	0	0	0	0	0	0
OC3	0	0	0	0	0	0	0	0	0	0	0	0	0	0
OC4	0	0	0	0	0	0	0	0	0	0	0	0	0	0
EC1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
EC2	0	0	0	0	0	0	0	0	0	0	0	0	0	0
EC3	0	0	0	0	0	0	0	0	0	0	0	0	0	0	Rank
A1	0.241	0.116	0.083	0.081	0.180	0.190	0.195	0.244	0.607	0.705	0.669	1	0	0	2
A2	0.547	0.622	0.727	0.746	0.084	0.084	0.084	0.094	0.256	0.190	0.243	0	1	0	1
A3	0.212	0.263	0.190	0.173	0.736	0.726	0.721	0.661	0.138	0.105	0.088	0	0	1	3

Source: authors’ own elaboration.

by taking the supermatrix to the power of “k”, where “k” is an arbitrary number. Table 10 is also designated as the limit matrix by some researchers. In this ongoing analysis, “k = 2” is considered, which means that the supermatrix has been squared to obtain the limit matrix shown in Table 10. This step provides the final priorities of the three AI alternatives considered in the network. This step takes into account the interdependencies and feedback among elements. It is to be noted that if the limit matrix is multiplied “x” times, it will yield the same results as Table 10.

Based on the final priorities of the three alternative choices evaluated in Table 10, rankings were prescribed in decreasing order, indicating the superior and inferior AI technologies in the list. The ANP is particularly useful for decision problems with complex interactions, such as strategic planning, resource allocation, and complex systems analysis. It allows decision-makers to capture and evaluate the intricate relationships among various elements, leading to more robust and comprehensive decision-making.

4.3. Data Analysis of TOPSIS

The TOPSIS method is an MCDM technique used to evaluate and rank alternatives based on multiple criteria. It helps decision-makers to make informed choices when faced with complex decision problems. The TOPSIS method follows a systematic approach to determine the best alternative by comparing their similarity to an ideal solution [56]. The subsequent steps may be followed for the successful implementation of the TOPSIS model:

Step 1 (define the decision problem): Clearly define the problem and identify the alternatives and criteria involved, as depicted in the hierarchical network in Figure 5. Alternatives are the different options or choices available, and criteria are the factors or attributes used to evaluate the alternatives. In this case, three alternative AI technologies were considered, and they were evaluated based on 10 potential factors.

Step 2 (construct a decision matrix): Create a decision matrix (R = m_i $\times$ n_j) according to Equation (16) that represents the performance of each alternative (r_ij) with respect to each criterion. The performance degree of each alternative was already calculated in the previous ANP stage, as evident from the limit matrix in Table 10. The matrix has “m” rows for alternatives and “n” columns for criteria. Fill the matrix with corresponding values that measure the performance of each alternative on each criterion, as shown in

Table 11. Performance evaluation matrix for TOPSIS.

Nature	Min	Max	Max	Max	Max	Max	Min	Max	Max	Min
Alternatives	T1	T2	T3	O1	O2	O3	O4	E1	E2	E3
A1	0.116	0.083	0.081	0.180	0.190	0.195	0.244	0.607	0.705	0.669
A2	0.622	0.727	0.746	0.084	0.084	0.084	0.094	0.256	0.190	0.243
A3	0.263	0.190	0.173	0.736	0.726	0.721	0.661	0.138	0.105	0.088
Square sum	0.469	0.572	0.593	0.581	0.570	0.565	0.506	0.452	0.544	0.514
Square root	0.685	0.756	0.770	0.762	0.755	0.752	0.711	0.673	0.737	0.717

Source: authors’ own elaboration.

. The nature of each criterion is also clearly indicated in Table 11, which shows that 7 out of the 10 factors are beneficial, whereas 3 are non-beneficial in nature.

$$\mathrm{R} ({\mathrm{m}}_{\mathrm{i}}\times {\mathrm{n}}_{\mathrm{j}})=\left[\begin{array}{c}{\mathrm{r}}_{11}\\ {\mathrm{r}}_{21}\\ \dots \\ {\mathrm{r}}_{\mathrm{i}1}\end{array} \begin{array}{c}{\mathrm{r}}_{12}\\ {\mathrm{r}}_{22}\\ \dots \\ {\mathrm{r}}_{\mathrm{i}2}\end{array} \begin{array}{c}\dots \\ \dots \\ \dots \\ \dots \end{array} \begin{array}{c}{\mathrm{r}}_{1\mathrm{j}}\\ {\mathrm{r}}_{2\mathrm{j}}\\ \dots \\ {\mathrm{r}}_{\mathrm{ij}}\end{array}\right]$$

(16)

Step 3 (normalize the decision matrix): Normalize the values in the decision matrix using Equation (17) to eliminate any scale differences among the criteria. TOPSIS follows a vector normalization procedure, and the normalized matrix is shown in

Table 12. Normalized matrix for TOPSIS.

Weights (wj)	0.071	0.448	0.161	0.004	0.052	0.014	0.007	0.023	0.160	0.061
Alternatives	T1	T2	T3	O1	O2	O3	O4	E1	E2	E3
A1	0.169	0.110	0.105	0.236	0.252	0.259	0.344	0.902	0.956	0.933
A2	0.908	0.962	0.969	0.111	0.111	0.111	0.133	0.380	0.258	0.339
A3	0.383	0.251	0.225	0.966	0.961	0.959	0.930	0.205	0.142	0.123

Source: authors’ own elaboration.

. This step ensures that all criteria are on the same scale and avoids bias caused by different measurement units.

$${{\mathrm{N}}_{\mathrm{ij}}}^{\mathrm{TOPSIS}}=\frac{{\mathrm{r}}_{\mathrm{ij}}}{\sqrt{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{r}}_{\mathrm{ij}}^2}}$$

(17)

Step 4 (determine the weighted normalized decision matrix): Assign weights to the criteria to reflect their relative importance or significance. These weights represent the global weights derived from Table 8. The weights indicate the decision-maker’s preferences or priorities for each criterion. Multiply each normalized value in the decision matrix by its corresponding weight using Equation (18) to obtain the weighted normalized decision matrix shown in

Table 13. Weighted normalized matrix for TOPSIS.

Alternatives	T1	T2	T3	O1	O2	O3	O4	E1	E2	E3
A1	0.012	0.049	0.017	0.001	0.013	0.004	0.002	0.021	0.153	0.057
A2	0.064	0.431	0.156	0.000	0.006	0.002	0.001	0.009	0.041	0.021
A3	0.027	0.112	0.036	0.004	0.050	0.013	0.006	0.005	0.023	0.007
Ideal best (IBj)	0.012	0.431	0.156	0.004	0.050	0.013	0.001	0.021	0.153	0.007
Ideal worst (IWj)	0.064	0.049	0.017	0.000	0.006	0.002	0.006	0.005	0.023	0.057

Source: authors’ own elaboration.

.

$${{\mathrm{WN}}_{\mathrm{ij}}}^{\mathrm{TOPSIS}}={{\mathrm{N}}_{\mathrm{ij}}}^{\mathrm{TOPSIS}} \times {\mathrm{w}}_{\mathrm{j}}$$

(18)

Step 5 (determine the positive ideal solution (PIS) and negative ideal solution (NIS)): Identify the PIS (S_i⁺) and NIS (S_i⁻) for each criterion. The PIS represents the best possible value for each criterion, while the NIS represents the worst possible value. For the PIS, the Euclidean distance measures the similarity or proximity of each alternative to the IB for each criterion. It is computed using Equation (19) by comparing the values of each alternative to the corresponding values of the best solution indicated in Table 13. For the NIS, the Euclidean distance measures the similarity or proximity of each alternative to the IW for each criterion. It is computed using Equation (20) by comparing the values of each alternative to the corresponding values of the worst solution indicated in Table 13. However, the PIS and NIS of each alternative are calculated and shown in

Table 14. Rating of alternatives using TOPSIS.

Alternatives	S+	S−	RCC	%	Rank
A1	0.411	0.142	0.256	25.635	2
A2	0.133	0.408	0.754	75.4413	1
A3	0.365	0.101	0.217	21.7158	3

Source: authors’ own elaboration.

.

$${{\mathrm{S}}_{\mathrm{i}}}^{+}=\sqrt{{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{({\mathrm{WN}}_{\mathrm{ij}}^{\mathrm{TOPSIS}}-{ \mathrm{IB}}_{\mathrm{j}})}^2}$$

(19)

$${{\mathrm{S}}_{\mathrm{i}}}^{-}=\sqrt{{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{({\mathrm{WN}}_{\mathrm{ij}}^{\mathrm{TOPSIS}}-{ \mathrm{IW}}_{\mathrm{j}})}^2}$$

(20)

Step 6 (calculate the relative closeness to the ideal solution): Calculate the relative closeness coefficient (RCC) for each alternative using Equation (21) by dividing the distance to the NIS by the sum of the distances to both the PIS and NIS. This step determines how close each alternative is to the ideal solution relative to the other alternatives. The RCC values of each alternative are presented in Table 14.

$${\mathrm{RCC}}_{\mathrm{i}}=\frac{{\mathrm{S}}_{\mathrm{i}}^{-}}{{\mathrm{S}}_{\mathrm{i}}^{+}+{ \mathrm{S}}_{\mathrm{i}}^{-}}$$

(21)

Step 7 (rank the alternatives): Rank the alternatives based on their RCC values, as portrayed in Table 14. The alternative with the highest RCC value is considered to be the most preferred or best alternative.

By following these steps, the TOPSIS method provides a systematic and quantitative approach to evaluate and rank alternatives based on multiple criteria, helping decision-makers to make more informed and rational decisions.

5. Results and Discussion

The preceding sections performed the mathematical computations for the MCDM tools employed. This section focuses on the main outcomes produced from the hybrid MCDM system deployed. The following section highlights the study’s primary conclusions, which are intended to identify the major hurdles to deploying AI technologies in the construction environment. This section presents the study results obtained from the fuzzy integrated hybrid system in a systematic manner, and their implications are thoroughly examined. The ongoing analysis was started with the Delphi managerial session, where 13 potential AI factors were initially examined, out of which 10 factors were found to be suitable for the current problem. These 10 parameters were finally qualified for the second ANP stage, followed by TOPSIS. We begin by outlining the key findings of the ANP and TOPSIS, providing a clear and concise overview of the findings.

5.1. Outcomes from the ANP

Let us start with the ANP analysis, which helped us to compute the weights of each of the primary and sub-criteria. Furthermore, for each pairwise comparison, all consistencies were evaluated to validate the experts’ choices. Table 8 presents the results of the ANP analysis. Table 8 also illustrates that the weightages derived from the ANP analysis are used to determine the rankings of the major factors and sub-factors. The components with the largest weightages are considered to be the most important, and the ranking order is determined by decreasing weight value. Among the key variables, technological hurdles have the highest weight contribution of 67.9%, making them the most important component in the category. It is also true that the construction industry relies heavily on cutting-edge technologies, which predominantly impact the AI systems of a sector. As a result, technological obstacles are among the most important factors impacting the construction industry’s AI infrastructure. Similarly, environmental factors are rated second, with a weight contribution of 24.4%, showing that they are another significant component in the list. Almost every organization today is concerned about the environment and strives to promote sustainability through their services. Finally, organizational criteria have the lowest weight contribution of 7.7%, making this the group’s least important primary element. Figure 6 shows a pie chart depicting the primary criteria contributions towards the analysis, in percentages.

Table 8 illustrates that the sub-factors within each major factor were allocated ranks based on the sub-criteria’s local weights (LWs) created by the ANP. The local rank (LR) of the sub-criteria reflects the most excellent and inferior attributes within each main factor category. Another thing to note is that the CR values in Table 8 demonstrate that the inconsistencies connected in each case are far lower than the upper bound limit of 10%. As a result, all of the pairwise comparisons involved are consistent and stable. Table 8 also gives the global weights (GWs) for each sub-criterion, supported by their global rank (GR), which helps the expert members to create an overview of the most significant criteria on an overall basis. Overall, the parameter with the highest weight value has the largest influence on the AI architecture, while the parameters with the lowest weight have a comparatively lower influence. The LW, GW, LR, and GR are represented with the help of column diagrams in Figure 7.

The GW selects the best parameter in the group and its contribution to the analysis. Furthermore, a GR is also prescribed, indicating that the value of data and algorithms (TC2), followed by advancements in innovations for traditional problems (TC3) and trust between different companies/copyright/ownership (EC2), are the significant factors in the list that most influence an organization’s AI environment. On the other hand, government/management (OC1), risk-taking ability (OC4), and information exchange and communication/interoperability (OC3) occupy the last three rank positions, indicating that they are the least important factors among the group and may contribute only marginally to achieving the objectives. Both LW and GW have their own significance; LW helps to recognize the value of each sub-factor within each key criteria category, whilst GW helps discover the most significant aspects that influence the entire analysis globally. Figure 8 depicts the primary criteria weights as well as the percentage contribution of each sub-factor within its corresponding main criterion group.

In addition to assessing the parametric weights, the ANP also aids in prioritizing the three AI solutions provided in Table 10. The limit matrix in Table 10 clearly illustrates that the priority value of the second alternative (A2) with respect to the goal is the highest, i.e., 0.547, followed by A1 (0.241) and A3 (0.212), and the preference rating order is shown in Table 10. As we can see, the ANP analysis achieves all of its objectives and successfully handles the weightage computation phase. The ANP was largely responsible for assessing the parametric importance and sorting out the priority rankings from highest to lowest. The ANP also supports decision-makers in determining the contribution and influence of each study parameter. As a result, the employment of the Delphi method, ANP, and TOPSIS has been used for a variety of purposes to achieve specific goals. Each technique employed in this study is significant and contributes to the overall study. In the next subsection, we focus on the goals achieved using the TOPSIS model.

5.2. Outcomes from TOPSIS

The outcome of the TOPSIS method provides rankings to the list of alternatives based on their relative closeness to the ideal solution, as can be seen in Table 14. The primary outcome of TOPSIS is the ranking of alternatives, which provides decision-makers with a clear knowledge of which option is the most desirable or beneficial based on the factors evaluated. This rating aids in decision-making by decreasing complexity and allowing for systematic assessment of options. Alternative A2 has the highest RCC value and, thus, is deemed to be the most favored or best alternative, whilst alternative A3 has the lowest value and, thus, is considered to be unfavorable. However, both the ANP and TOPSIS yield the same preference order, with A2 being the best option and A3 being the worst. The TOPSIS technique’s results give decision-makers a clear picture of the relative performance of the options, allowing them to make intelligent judgments. The ranked list enables decision-makers to identify the best-ranked alternative(s) based on the criteria and their preferences. Furthermore, the outcomes from the TOPSIS method can provide insights into the relative strengths and weaknesses of each solution. Decision-makers can identify which alternatives perform well on specific criteria and which fall short by comparing the distances to the PIS and NIS. These data can be useful for additional analysis and decision-making. It is vital to remember that the TOPSIS approach produces results based on the input data and the weights provided to the criteria. The TOPSIS method yields a systematic and objective examination of alternatives, assisting decision-makers in picking the best option based on various factors. Figure 9 depicts the alternatives’ ranking graphically using a line diagram.

6. Conclusions

This research paper aimed to investigate the role of AI technologies in the construction industry by employing the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment. Through an extensive analysis, the study shed light on the significance and implications of AI’s adoption in the construction sector. The findings of this research indicate that AI technologies have a profound impact on various aspects of the construction industry. The Delphi-ANP-TOPSIS hybrid MCDM approach, incorporating expert opinions and fuzzy logic, provided a robust framework for evaluating and prioritizing AI technologies based on multiple criteria. This framework facilitated informed decision-making, enabling stakeholders to identify the most suitable AI solutions for different applications within the construction industry. The study highlighted the key sub-criteria within the selected criteria, including cost/budget, employee workforce, information exchange and communication/interoperability, environment, upstream and downstream policies/laws, trust between different companies/copyright/ownership, and social impacts. Each sub-criterion played a crucial role in assessing the impact of AI technologies on the construction industry and identifying opportunities for the improvement and mitigation of potential challenges.

This research emphasizes the importance of considering the environmental implications of AI’s implementation in the construction sector. It emphasizes compliance with environmental impact assessment regulations, waste management, sustainability policies, and supply-chain compliance as critical factors for the responsible and sustainable adoption of AI. Additionally, the legal and ethical considerations surrounding AI technologies, such as copyright, ownership, and trust between different companies, were explored. Addressing these concerns is vital for ensuring fair and transparent AI practices and fostering a collaborative and trustworthy environment among industry participants. This research paper also underscores the social impacts of AI technologies in the construction industry. Workforce displacement, worker safety and health, stakeholder engagement, inclusivity, and the digital divide were highlighted as key areas requiring attention. By addressing these social factors, organizations can maximize the positive impacts of AI technologies while mitigating potential negative consequences on workers, communities, and society as a whole.

Overall, this research paper contributes to the existing body of knowledge by providing insights into the role of AI technologies in the construction industry under a fuzzy environment. The utilization of the Delphi-ANP-TOPSIS hybrid MCDM approach allowed for a comprehensive and systematic evaluation of AI technologies, enabling stakeholders to make informed decisions and prioritize the most suitable solutions. The identified sub-criteria within various criteria emphasized the significance of considering cost, workforce, environmental impact, legal and ethical aspects, and social implications in the adoption and implementation of AI technologies. As AI continues to evolve and shape the construction industry, future research should focus on monitoring and evaluating the long-term effects and benefits of AI’s implementation, refining the MCDM framework, and exploring emerging technologies and innovative applications that can further enhance the efficiency, sustainability, and social impact of the construction industry in the era of AI.

6.1. Theoretical Contributions

The theoretical contribution of investigating the role of AI technologies in the construction industry using the Delphi-ANP-TOPSIS hybrid multi-criteria decision-making (MCDM) concept under a fuzzy environment can be outlined as follows:

• Integration of MCDM methods: The theoretical contribution lies in the integration of various MCDM techniques (i.e., the Delphi method, ANP, and TOPSIS) within a hybrid framework. This integration allows for a comprehensive analysis and evaluation of the role of AI technologies in the construction industry. By combining these methods, this study was able to capture and weigh multiple criteria, considering the complex relationships and dependencies among them.
• Incorporation of a fuzzy environment: The theoretical contribution also lies in considering a fuzzy environment within the MCDM framework. Fuzzy logic allows for handling uncertainties and vagueness associated with subjective judgments and imprecise data in decision-making processes. By incorporating fuzzy sets and fuzzy numbers, this study was able to effectively handle and model the inherent uncertainties involved in evaluating the role of AI technologies in the construction industry.
• Exploration of AI technologies in construction: This investigation focused specifically on the role of AI technologies in the construction industry. By utilizing the hybrid MCDM framework, this study aimed to identify and evaluate the potential impact, benefits, and challenges associated with the adoption of AI in construction. This contributes to the theoretical understanding of how AI can transform and enhance various aspects of construction processes, including project management, resource allocation, risk management, and decision-making.
• Insights for decision-makers: Another contribution may be highlighted as providing valuable insights and guidance for decision-makers in the construction industry. By utilizing the hybrid MCDM framework, this study generated rankings, priorities, and recommendations for AI technologies based on the identified criteria and their relative importance. This information can assist decision-makers in making informed choices regarding the adoption and implementation of AI technologies in construction, considering the fuzzy and uncertain nature of the decision environment.

The theoretical contribution of investigating the role of AI technologies in the construction industry using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment lies in the integration of MCDM methods, the incorporation of fuzzy logic, the exploration of AI technologies in construction, and the provision of valuable insights for decision-makers. This research contributes to advancing the understanding and application of AI in the construction industry, addressing the complexities and uncertainties involved in decision-making processes.

6.2. Managerial Implications

This research paper, investigating the role of AI technologies in the construction industry using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment, offers several practical implications for industry professionals, policymakers, and researchers. These implications are derived from the study’s findings and can guide stakeholders in making informed decisions regarding the adoption of AI in the construction sector. Below are some practical implications.

• The Delphi-ANP-TOPSIS hybrid MCDM idea provides a viable decision-support framework in this research article. This approach can be used by industry professionals to evaluate and prioritize different AI technologies based on a variety of variables, including cost, workforce, environmental effect, legal and ethical issues, and social consequences. This framework facilitates informed decision-making and aids in the selection of the best AI solutions for specific construction industry applications.
• The conclusions of this research emphasize the need to address the technological and environmental consequences of AI’s implementation. Stakeholders can efficiently allocate resources by selecting cost-effective AI technologies that deliver the most benefits in terms of productivity, efficiency, and sustainability by applying the proposed hybrid MCDM paradigm. This can result in better resource usage and project outcomes.
• This study underlines the importance of assessing the environmental impact of artificial intelligence technology in the construction industry. Stakeholders can contribute to environmental sustainability by adhering to environmental impact assessment standards, implementing sustainable practices, and taking into account waste management and supply-chain compliance criteria. This indicates that when selecting and deploying AI technology, green building standards, energy efficiency certifications, and other sustainability frameworks must be considered.
• This study emphasizes the importance of resolving the legal and ethical issues of AI’s deployment. By establishing unambiguous agreements and collaborations, stakeholders can ensure compliance with copyright and ownership requirements while also fostering trust between diverse companies. Organizations can prevent legal issues, preserve intellectual property rights, and build a collaborative and trusting environment among industry participants.
• This study emphasizes the importance of considering societal implications and including stakeholders throughout the AI adoption process. This includes addressing labor displacement concerns by providing chances for reskilling or upskilling, as well as guaranteeing worker safety and health. Involving varied stakeholders and communities in decision-making processes can also lead to more inclusive and socially responsible AI implementation.
• This study emphasizes the significance of ongoing monitoring and evaluation of AI technology in the construction industry. Researchers can improve the framework by building on the proposed Delphi-ANP-TOPSIS hybrid MCDM approach. Furthermore, investigating future AI technologies and novel applications can provide new opportunities for enhancing efficiency, sustainability, and social impact in the construction industry.

These practical implications provide guidance for industry professionals, policymakers, and researchers aiming to integrate AI technologies effectively and responsibly in the construction industry. By considering the decision-support framework, resource allocation, environmental sustainability, legal and ethical compliance, social impacts, and ongoing research, stakeholders can maximize the benefits of AI’s adoption while minimizing potential risks and challenges.

6.3. Limitations

Although the Delphi-ANP-TOPSIS hybrid MCDM paradigm in a fuzzy environment gives significant insights, the authors encountered some constraints that need to be acknowledged while investigating the use of AI technologies in the construction sector. These limitations indicate areas that require additional research and development. Below are some of the study’s drawbacks.

• Data availability and quality: One potential drawback is the availability and quality of study data. The Delphi-ANP-TOPSIS hybrid MCDM concept’s effectiveness and accuracy are strongly reliant on the input data provided by experts or stakeholders. If the data are restricted, partial, or subjective, this may impact the reliability and validity of the findings. This restriction could be addressed in future studies by conducting more extensive data-gathering activities and using more comprehensive and standardized datasets. Moreover, gathering essential data for Delphi-ANP-TOPSIS might be a difficult task. It can be challenging to obtain accurate and reliable information to quantify the interactions and dependencies between parts, especially when dealing with subjective or intangible issues. Data collection efforts may also necessitate a large amount of time and resources.
• Biasedness: The Delphi method relies on expert input and judgments, which might add bias and subjectivity to the decision-making. Personal viewpoints, knowledge constraints, or individual biases may influence expert opinions, affecting the credibility and objectivity of the outcomes. The ANP also depends on expert assessments and pairwise comparisons to define the comparisons and priorities among criteria and alternatives. These decisions are frequently subjective and sensitive to biasedness, which can add uncertainty and impair the accuracy of the outcomes. The accuracy of pairwise comparisons is strongly dependent on the decision-makers’ skill and knowledge.
• Complexity: The ANP is concerned with the modeling of complicated decision issues that have interdependencies between criteria and options. The process of building a network and defining relationships between pieces can be complex and time-consuming. Managing and analyzing vast networks with multiple constituents might add to the decision-making process’s complexity.
• Pairwise comparison difficulties: The ANP uses pairwise comparisons to determine the relative relevance of components. It can be difficult to perform these comparisons effectively and consistently, especially when dealing with a high number of elements. Decision-makers may have difficulty assigning exact and consistent values to reflect the relative importance accurately.
• Lack of diversity: The Delphi technique is often used with a small group of experts or stakeholders. While this is beneficial in terms of acquiring specialized knowledge and skills, it can also lead to a lack of diversity in perspectives. Because of the inadequate representation of perspectives, the range of choices or criteria considered may be limited, perhaps disregarding key ideas or insights.
• Time- and resource-intensive: The Delphi technique frequently necessitates several iterations of data collection and feedback. This iterative procedure can be time-consuming, necessitating substantial resources for assembling expert panels, disseminating and evaluating questionnaires, and summarizing data. The extended timeframe may not necessarily coincide with project deadlines or the urgency of decision-making.
• Difficulty in quantification: The Delphi method is a largely qualitative approach that is used to collect expert opinions via organized surveys or interviews. Quantifying and integrating these qualitative inputs into a quantitative MCDM framework may be difficult. Converting qualitative data to quantitative measurements, such as criterion weights or ratings, may contribute more uncertainty and subjectivity.
• Fuzziness and subjectivity: The fuzzy environment used in this research introduces subjectivity and ambiguity. The employment of linguistic concepts and fuzzy logic in decision-making provides inherent ambiguity, which might change depending on interpretation. While the fuzzy environment provides for more flexible decision-making, it also increases the complexity in capturing the problems effectively and measuring the fuzzy information. Future research could explore ways to reduce subjectivity and improve the robustness of the fuzzy environment in the MCDM paradigm.

Recognizing these limitations will allow future studies to fill gaps and improve the role of AI technologies in the construction industry. Overcoming these constraints will lead to more robust decision-making frameworks and responsible integration of AI technologies in the construction sector.

6.4. Future Work

This research paper, investigating the role of AI technologies in the construction industry using the Delphi-ANP-TOPSIS hybrid MCDM concept under a fuzzy environment, opens up several avenues for future work. These areas of research could further enhance the understanding and application of AI technologies in the construction sector. Below are some potential directions for future research.

• Several criteria and sub-criteria have been identified in this research article for evaluating the use of AI technologies in the construction industry. Future research could look into other criteria that are relevant to different construction environments or innovative AI applications. For example, criteria pertaining to safety, risk management, or project scheduling could be integrated to create a more comprehensive evaluation framework.
• The Delphi-ANP-TOPSIS hybrid MCDM framework was proposed in this study for decision-making. Future studies could concentrate on refining and upgrading the hybrid model by including other decision-making approaches or investigating alternative MCDM frameworks and comparing the findings to the current results. The goal would be to improve the decision-making process’s accuracy, robustness, and applicability.
• The present study only evaluated three AI possibilities, and they were prioritized accordingly. However, more AI alternatives may be added in the future to broaden the scope of choice prioritization.
• This research was primarily concerned with assessing the role of AI technologies in the construction industry from a short-term perspective. Future research could look into the long-term effects of AI’s adoption, taking into account issues like project performance, productivity, sustainability, and overall industrial transformation. Longitudinal studies could provide insights into how AI technologies evolve and how they affect the construction industry over time.
• Comparative research across industries or geographical regions can provide useful insights into the particular difficulties and opportunities connected with AI’s adoption in the construction industry. Benchmarking AI deployment processes can aid in the identification of best practices, success factors, and areas for improvement.

By focusing on these areas for future work, researchers can further advance the understanding and practical application of AI technologies in the construction industry. This will lead to more informed decision-making, improved project outcomes, and sustainable development in the era of AI.