1. Introduction and Background
In our increasingly urbanizing world, the construction industry is not only vital for the development of cities but also for the global economy, accounting for approximately 13% of the world’s gross domestic product (GDP) [
1]. This estimate is based on a combination of economic data, industry reports, and various sources of information to estimate the percentage. Despite its urbanization and economic significance, the industry faces a range of challenges that undermine productivity and efficiency, including cost overruns, project delays, and quality issues. This is evident in a Mckinsey 2020 report that indicated that large construction projects tend to exceed their schedules by 20% and exceed their budgets by 80% [
1]. This is compounded by a lack of digitalization, and the manual nature of the industry makes projects more complex and tedious [
2]. To combat these challenges and transform the construction sector, particularly in a sustainable way, the integration of artificial intelligence (AI) has emerged as a promising solution [
2,
3,
4].
The widely accepted definition of AI states that “AI is the study of how to make machines do things, which at the moment, people do better” [
5]. However, the industry has experienced sluggish productivity growth, falling behind other sectors with an annual rate of only 1% over the past two decades [
6,
7]. This slow growth can be attributed to inherent characteristics of the industry, such as cyclical demand, limited capital investment, and a lack of standardization. Furthermore, the industry’s inadequate investment in construction innovation and the prevalence of small, specialized subcontractors with limited technological advancements have hindered the widespread adoption of automation and constrained its overall growth potential [
7,
8,
9]. This is apparent in a recent study conducted by KPMG that indicated only 8% of construction companies are identified as highly innovative, with the remaining 92% categorized as moderate or low innovators [
4].
The history of AI in construction dates back several decades, with the concept evolving alongside advancements in technology and computing power [
10]. In the initial phases of AI integration in construction, three significant applications emerged. First, AI optimizes project schedules and planning for enhanced efficiency. Second, AI-enhanced risk management identifies and mitigates potential issues proactively. Finally, AI’s precise cost estimation helps prevent budget overruns, contributing to improved financial control [
5,
11].
The application of AI in construction involves the use of intelligent machines and programs that mimic human cognitive functions, such as learning, problem-solving, and decision-making. Furthermore, the major components of AI include leaning, knowledge representation, perception, planning, action, and communication [
5]. In the early stages, AI in construction focused primarily on automation, aiming to enhance efficiency and productivity on construction sites [
12]. As computing power increased and data analysis capabilities improved, the concept of AI expanded to include machine learning and data-driven decision-making. Machine-learning algorithms allow systems to analyze large amounts of data, recognize patterns, and make predictions or recommendations based on past experiences [
13]. This capability has been applied to various aspects of construction, including project planning, design optimization, risk management, and resource allocation [
14,
15].
In recent years, AI in construction has further evolved around the following major subfields; machine learning, computer vision, natural language processing, knowledge-based systems, optimization, robotics, and automated planning and scheduling [
5,
16]. These emerging technologies have enabled real-time data collection, optimized decision-making, and improved project management [
17,
18]. Today, deep learning, a subset of AI has gained significant attention in the construction industry due to its potential to address complex technical challenges. It involves training artificial neural networks with multiple layers to learn patterns and make accurate predictions based on large datasets [
13,
19,
20]. Deep learning offers a crucial advantage to the construction industry as it gives the ability to process intricate data, allowing for informed decisions, improved efficiency, and enhanced project outcomes [
2]. Furthermore, by analyzing data, deep learning models can predict hazards, identify defects, optimize scheduling, and enhance design decisions. Overall, deep learning has the potential to revolutionize construction processes [
10,
21,
22].
AI continues to emerge as a powerful tool to tackle productivity challenges in the construction industry and drive improvements in performance, efficiency, and innovation [
7,
23]. By simulating human cognitive functions, AI technologies offer opportunities to transform business models within construction [
24,
25]. Through AI adoption, construction operators can optimize resource allocation, automate tasks to address skill shortages and unlock higher levels of productivity and efficiency [
26]. Nonetheless, the practical implementation of AI in construction faces several obstacles. One key challenge is the need for accurate and comprehensive data to train AI algorithms, which can be costly and time-consuming for many construction companies [
27]. The dynamic outdoor environmental conditions and the non-standardized nature of building designs pose further complexities in effectively applying AI. As a result, while larger construction companies may reap some benefits from AI technologies, there remains a lack of widespread knowledge and established frameworks within the industry. This has led to ongoing debates regarding the future of the construction workforce and the potential impact of AI on jobs [
8,
28].
Although there have been numerous studies examining the application of AI in construction, it is important to note that many of these studies have been limited in their scope. These include small sample sizes, theoretical focus, limited long-term analysis, data quality issues, and concerns about bias. Addressing these constraints through interdisciplinary research is essential to gain a comprehensive understanding of AI’s potential in construction. They have focused on specific aspects of AI implementation in construction, often neglecting to comprehensively review research clusters related to the four pillars of sustainability and the different stages of construction projects [
11]. Sustainability in construction involves creating built environments that meet present needs without compromising the ability of future generations to meet their own needs. It combines environmental responsibility, social equity, economic viability, and responsible governance to create balanced and positive outcomes for society and the environment [
28]. The four pillars of sustainability are the following.
Economic sustainability: The ability of construction projects to create long-term economic value while considering the needs of both present and future generations. It involves ensuring the financial viability of construction projects, promoting fair trade practices, supporting local economies, and optimizing resource allocation [
5]. This pillar emphasizes the importance of cost-effectiveness, profitability, and the long-term economic benefits derived from sustainable construction practices [
29,
30].
Social sustainability: Meeting the needs and improving the quality of life for individuals and communities affected by construction projects. It encompasses factors such as social equity, inclusivity, health, safety, and well-being. Socially sustainable construction practices involve providing safe working conditions, promoting diversity and equal opportunities, respecting local cultures and traditions, and enhancing community engagement throughout the project lifecycle [
28].
Environmental sustainability: Minimizing the negative impact of construction activities on the natural environment. It involves practices that conserve resources, reduce pollution, promote energy efficiency, and mitigate climate change [
31]. Examples include using renewable energy sources, employing sustainable construction materials, reducing waste generation, and implementing effective water and land management strategies [
32].
Governance sustainability: Promoting transparency, accountability, and ethical decision-making in construction projects. It emphasizes the importance of effective governance structures, policies, and regulations to ensure compliance with environmental, social, and economic standards. This pillar focuses on promoting responsible practices, preventing corruption, establishing clear frameworks for decision-making, and fostering collaboration between stakeholders to achieve sustainable outcomes [
33].
The use of AI is evident across all four pillars and plays a vital role in achieving sustainability. Economically, AI leverages historical data for precise cost estimates, optimized resource allocation, and waste reduction [
34]. Socially, AI’s predominant applications lie in safety monitoring, community engagement, and workforce development, fostering safer sites, inclusive feedback analysis, and efficient training [
5]. Environmentally, AI optimizes energy use, promotes sustainability, and reduces waste through consumption analysis, improved systems, material evaluation, and waste reduction efforts [
34]. In governance, AI ensures compliance, transparency, and oversight by monitoring data for regulations, enhancing accountability, and refining project management and risk mitigation [
22].
Each of these pillars plays a crucial role in ensuring sustainable construction practices. However, previous studies often tend to focus on a subset of these dimensions or overlook their interconnectedness [
2]. By taking a holistic approach and considering all four pillars, a more comprehensive understanding of the potential impact of AI on sustainable construction can be achieved. Additionally, construction projects consist of various phases, including planning, design, construction, and operation and maintenance. Each phase presents unique challenges and opportunities for implementing AI. However, many studies have focused on specific phases or failed to consider the entire project lifecycle. Understanding the potential of AI across all phases of construction is vital for developing effective strategies and harnessing its full potential [
29,
35].
This study adopts a scientometric approach to analyze scholarly research published in the last two decades. It aims to bridge gaps in the existing literature by conducting a comprehensive analysis that encompasses the four pillars of sustainability and considers the complete lifecycle of construction projects. The study has the objectives to uncover historical and emerging trends, identify research clusters, and present a summary of influential authors, publications, countries, universities, and publishing sources in the domain of AI in construction. The primary contribution of this research lies in improving our understanding of the potential opportunities and challenges associated with AI in construction and providing practical insights for its application from commencement to completion of a project. Overcoming AI implementation challenges in construction involves education and training for professionals, collaboration with experts, and robust data management. Regulatory frameworks should also be established to ensure ethical deployment. These solutions enable effective AI adoption and innovation in the industry. By undertaking such research, the construction industry can develop more robust frameworks and strategies to effectively leverage the capabilities of AI and achieve sustainable and efficient construction practices.
Following this introduction, the paper is organized as follows.
Section 2 outlines the research methodology employed in the study.
Section 3 presents the results that cover a range of aspects including general observations, academic influence analysis, research clusters in AI in construction literature, historical clusters, and emerging trends in AI research specific to construction and, along with the analysis findings of research trends and clusters, considers both the four pillars of sustainability and the four construction phases (planning, design, construction, and operation and maintenance).
Section 4 offers a detailed discussion of the results, providing insights and interpretations.
Section 5 concludes the paper.
2. Methodology
This study undertakes a scientometric analysis of existing AI and construction literature to address the following research questions: (a) What are the different areas of focus in AI in construction research? (b) What are the historical clusters, emerging trends, research clusters, and the correlation between sustainability pillars and project phases in AI in construction? With the guidance of the key literature—e.g., [
36,
37,
38]—and using scientometric techniques, the study creates a knowledge connection map that visually represents qualitative data and helps to provide a clearer understanding of the research clusters [
39]. This approach can provide valuable insights and a deeper understanding of the topic under investigation [
40].
An extensive literature review was conducted to identify the existing applications of AI in the construction industry. The primary database search was performed on Scopus and was then validated by data from other databased such as the Institute of Electrical and Electronics Engineers (IEEE), Association for Computing Machinery (ACM), and Science Direct were used for validation. These databases were selected for their collection of high-impact publications in construction, engineering, and computer science. Scopus, being the largest citation database was chosen as the primary data source, while the others were utilized for downloading full articles and validating the data. Furthermore, Scopus provides the added advantage of allowing files to be exported in multiple formats that are compatible with mainstream scientometric analysis software—e.g., VOSviewer 1.6.19.
The search was performed using 25 keywords of the subfields and the construction industry: (“Artificial Intelligence*” OR “AI”) AND (“Construction Industry” OR “Building”) AND (“Sustainable” OR “Sustainability”) OR “Urban Environment” OR “Construction Projects” OR “Automation” OR “Machine Learning” OR “Deep Learning” OR “Robotics” OR “Industry 4.0” OR “Building 4.0” OR “Urban Development” OR “AI Application” OR “Construction Projects” OR “Green Buildings” OR “Building Information Modeling” OR “BIM” OR “Smart Cities” OR “Smart Buildings” OR “Technology” OR “Lifecycle” OR “Civil Construction” using advanced search to achieve the focus of this study.
The search task of literature data was conducted in July 2023 by covering the publication between 1 January 2000 to 8 July 2023—covering over two decades of publications on the topic. Excluding the small number of publications with information absence—i.e., undefined document type and authors—the search resulted in selected a total of 9564 publications from Scopus repositories, including conference papers, articles, conference reviews, book chapters, and gray literature. It is noted that journal papers are extended, structured pieces published in journals, while a conference paper is a concise, focused work presented at conferences. After extracting the papers from the database, they were carefully checked for completeness and screened to remove any publications that did not meet the inclusion criteria. The primary criteria for inclusion in this study were articles that described or evaluated practical applications of AI subfields and techniques in the construction industry, based on information from the abstract, title, or full-text article when needed. Data extracted from each article included the application area in construction, the methodology/techniques used, and the findings.
As a result of this screening process, a final selection of 3710 publications was considered relevant and included for further investigation. From these screened papers, the publications were sorted based on the author keywords and abstract into the four pillars of sustainability, namely: (a) economic; (b) social; (c) environmental; and (d) governance, as well as into different project phases, such as: (a) planning; (b) design; (c) construction; and (d) operation and maintenance. After the search task was completed, full records of the resulting publications, including citation and bibliographical information, abstracts, keywords, funding details, and other relevant data were exported in CSV format. This format was chosen to ensure compatibility with the selected data analysis software, namely VOSviewer [
3].
VOSviewer is a widely used software tool designed to visualize and analyze bibliometric networks. Allowing researchers to comprehend intricate relationships and trends within the extensive scholarly literature. By processing bibliographic data encompassing publication records, citations, and keywords, VOSviewer creates visual representations that unveil connections and patterns among different elements in the dataset [
3]. It achieves this by constructing networks of nodes (authors, keywords, journals) and edges (co-citation or co-occurrence relationships), arranging them in the visualization space based on their associations. This interactive visual map facilitates the identification of key authors, influential papers, collaboration trends, and emerging research areas [
4]. VOSviewer has become a popular tool for scientometric research in various fields, including thermal comfort and building control [
41], sustainable urban development [
42], circularity in construction [
43], autonomous vehicles [
44], and smart homes [
45]—just to name a few. In this study, VOSviewer was used to analyze AI in construction literature and create a range of visual diagrams, such as co-authorship network maps, citation-based network maps, and co-occurrence network maps, to help visualize and qualitatively analyze the literature data [
39].
To ensure the reliability and validity of the analysis, repeated validations of the results were conducted, including duplicate screening of initial data, retesting of the software, and random selective tests of outputs. Scientometric analysis, while valuable for assessing research trends and patterns, comes with inherent limitations that should be considered [
40]. One significant constraint is its heavy reliance on bibliometric data, primarily citations and publication counts. Such data might not provide a comprehensive picture of research quality, impact, or context. Another limitation is not accounting for qualitative aspects of research [
41]. Consequently, this may lead to the inclusion of studies that have garnered citations due to factors other than their genuine impact or contributions. Lastly, scientometric analysis focuses on academic impacts and may overlook non-academic impacts un which have significant real-world implications or application. In essence, these limitations can lead to a distorted view of research trends, impact, and relevance. To mitigate these effects, it is crucial to acknowledge these limitations and complement scientometric analyses with qualitative evaluations and a broader consideration of research context and influence [
2].
4. Findings and Discussion
This study conducted a scientometric analysis from the sustainability and construction phases lenses to map over two decades of AI in construction research. The study analyzed a total of 3710 literature pieces published between January 2000 and February 2023 (spanning over two decades), intending to gain a clearer understanding of the historical clusters and research clusters in construction, as well as in the context of the four pillars of sustainability and the four construction phases.
The scientometric analysis disclosed that: (a) Literature on AI in construction has experienced steady growth during the last two decades; (b) Machine learning, deep learning, and big data are seen as the key enabling digital technologies; (c) Economic and governance pillars of sustainability exhibit the highest potential for AI adoption; (d) Design and construction phases demonstrate substantial advantages for AI adoption; (e) AI is, despite adoption challenges, a strong driver of the construction industry modernization; (f) By incorporating AI, the construction industry can advance towards a more sustainable future by consolidating its processes.
The identified research clusters in AI within the construction domain encompass various areas, including automation, digital twin, big data, deep learning, machine learning, information systems, and simulation. These clusters have expanded the research scope beyond their primary focus and have led to the emergence of new directions in the field. The subsequent overview presents the key findings based on the key research clusters (see
Table 10,
Table 11 and
Table 12).
Based on these research clusters, the key opportunities and challenges of each construction phase and sustainability pillar can be identified.
The study’s findings hold significant implications with actionable recommendations for the construction industry. First, it is imperative to recognize the steady growth of AI literature within the construction realm over the span of the last two decades. This awareness can serve as a vital starting point, prompting industry stakeholders to remain vigilant and responsive to the evolving landscape of AI advancements. By staying informed about the latest AI trends and research developments, the industry can proactively position itself for growth and innovation [
7,
8].
Second, an emphasis on prioritizing the integration of key enabling technologies is paramount. Machine learning, deep learning, and big data have emerged as foundational pillars for driving effective digital transformation. To navigate the complexities of modern construction challenges, embracing these technologies can empower the industry with data-driven insights, predictive capabilities, and enhanced decision-making [
50]. The following are key opportunities that AI can provide to the construction industry:
Design Complexity and Optimization: Modern construction projects involve complex designs that must be optimized for various factors, including structural stability, energy efficiency, and cost-effectiveness. AI-powered algorithms can analyze countless design variations to identify optimal solutions quickly, enhancing design efficiency and accuracy [
41].
Project Planning and Scheduling: The complexity of construction project schedules often leads to delays, resource conflicts, and cost overruns. AI can analyze historical project data, real-time progress, and external factors to generate dynamic and adaptable schedules that account for uncertainties and potential disruptions [
45].
Risk Assessment and Management: The construction industry has constant uncertainties that can lead to project risks. AI’s predictive analytics can forecast potential risks by analyzing historical data and project parameters, enabling proactive risk mitigation strategies and better-informed decision-making [
33].
Quality Control and Defect Detection: Ensuring the quality of construction work is a persistent challenge. AI-powered visual recognition systems can detect defects, discrepancies, and deviations from design plans by comparing real-time construction progress to digital models, ensuring adherence to specifications and standards [
48].
Resource Allocation and Management: Efficiently allocating labor, materials, and equipment is vital for project success. AI can optimize resource allocation by analyzing project requirements, availability, and constraints, thus minimizing waste, and enhancing resource utilization [
2].
Safety Monitoring and Compliance: Safety concerns are paramount in construction. AI-driven sensors, cameras, and wearable devices can monitor work environments for potential safety hazards in real time, alerting workers, and supervisors to risks and ensuring compliance with safety regulations [
35].
Data Integration and Collaboration: Construction projects involve multiple stakeholders, each contributing diverse data sources. AI can facilitate seamless data integration, enabling improved collaboration among project teams and decision-makers by providing a unified platform for information sharing and analysis [
47].
Supply Chain Optimization: Managing the supply chain efficiently to ensure timely delivery of materials and resources is a significant challenge. AI algorithms can predict demand, optimize procurement, and monitor logistics to prevent disruptions and delays [
52].
Environmental Impact Mitigation: Sustainable construction practices are essential for minimizing the industry’s environmental footprint. AI can assess and model the environmental impact of construction activities, suggesting eco-friendly materials, energy-efficient designs, and waste reduction strategies [
10].
Post-Construction Maintenance and Operations: Maintaining and operating constructed assets efficiently is an ongoing challenge. AI-powered predictive maintenance algorithms can analyze real-time data from sensors to anticipate maintenance needs and optimize asset performance [
26].
The third key recommendation centers on the strategic allocation of efforts toward the economic and governance pillars of sustainability. The study’s findings underscore the substantial potential for AI adoption in these areas, showcasing its capacity to streamline operations, optimize resource allocation, and strengthen compliance measures. This strategic alignment between AI and sustainability objectives can foster improved economic outcomes while concurrently bolstering social and governance standards [
52]. Furthermore, tapping into the substantial advantages outlined in AI adoption, particularly within the design and construction phases, can drive substantial performance improvements. Lastly, by leveraging AI-powered tools for efficient project planning, risk assessment, and real-time monitoring, the industry can elevate project execution while minimizing errors and delays.
Amid the challenges mentioned, the integration of AI into construction and the adoption of sustainable practices across the pillars of sustainability present noteworthy opportunities for the industry. Embracing these technologies and principles could catalyze transformative shifts in project planning, design, construction, and management [
15]. The advantages of AI, encompassing automation, data-driven decision-making, and advanced analytics, could yield enhanced project efficiency, cost reduction, and heightened productivity. In parallel, the adoption of sustainable practices might drive resource optimization, minimize environmental impact, and improve social outcomes. By proactively addressing challenges and capitalizing on the potential of AI and sustainability, the construction industry can steer itself toward a more sustainable trajectory. This approach aligns projects not only with economic viability but also with environmental and social responsibility [
5]. This cohesive strategy contributes to the broader goals of sustainable development, paving the way for more sustainable urban futures [
29,
61].
5. Conclusions
This study presents a comprehensive scientometric analysis of 3710 published papers on AI in construction over the past two decades with a particular angle from sustainability and construction phases. By examining the existing literature, this study provides an updated and concise overview of the field’s knowledge structure and evolution. The findings reveal a progression from basic automation to more advanced neural network platforms, with a strong focus on machine learning, deep learning, big data, and IoT. The concept of AI in construction has expanded over time, encompassing emerging technologies such as natural language processing, virtual reality, and augmented reality [
41]. The integration of intelligent systems and algorithms has shown great potential in improving productivity, efficiency, safety, and sustainability within the construction industry [
62]. As technology continues to advance, AI is expected to play an even larger role in driving innovation and transforming traditional construction practices [
5,
29,
46,
63].
Additionally, this comprehensive analysis of the integration of the pillars of sustainability into the various construction phases highlights the benefits, challenges, and opportunities associated with each phase. By identifying the specific contributions and implications of AI technologies in conjunction with sustainability principles, this research provides valuable insights for industry professionals, policymakers, and researchers seeking to drive sustainable transformations in the construction industry. The synthesized information on the benefits, challenges, and opportunities serves as a foundation for informed decision-making and strategic planning in implementing AI-driven solutions while considering sustainability goals. This study provides valuable insights into the present state of AI in construction, paves the way for future research and development, and underscores the significance of sustainable practices in shaping the industry’s future.
Although this research provides a sound foundation, several areas warrant further investigation. Future research can focus on addressing the identified challenges, such as the cost-effectiveness of implementing AI technologies in the construction industry, ensuring equitable access to AI-driven solutions, and resolving issues related to data integration and interoperability. Moreover, it should delve into the legal consequences that AI may introduce concerning intellectual property ownership, liability, data privacy, standards, ethics, contracts, cybersecurity, employment impact, transparency, and the evolving legal landscape. Additionally, it should investigate the potential societal and ethical impacts of AI adoption in construction, encompassing employment effects and the role of human workers in an increasingly automated setting. Further studies could focus on enhancing AI algorithms and models for more precise and streamlined decision-making in sustainability-related domains. Furthermore, the formulation of comprehensive frameworks and guidelines for seamlessly integrating AI technologies with sustainability principles throughout various construction phases would prove advantageous. By pursuing these research avenues, scholars can continue advancing the field, paving the way for future construction practices that are both sustainable and efficient.