The Relationship Between Artificial Intelligence (AI) and Building Information Modeling (BIM) Technologies for Sustainable Building in the Context of Smart Cities

Abstract

The development of information technologies has been exponentially applied to the architecture, engineering, and construction (AEC) industries. The extent of the literature reveals that the two most pertinent technologies are building information modeling (BIM) and artificial intelligence (AI) technologies. The radical digitization of the AEC industry, enabled by BIM and AI, has contributed to the emergence of “smart cities”, which uses information technology to improve urban operational and sustainable efficiency. Few studies have investigated the roles of AI and BIM in AEC from the perspective of sustainable buildings in assisting designers to make sustainable decisions at building and city levels. Therefore, the purpose of this paper is to explore the research status and future development trends in the relationship between AI and BIM-aided sustainable building in the context of the smart city to provide researchers, designers, and technology developers with potential research directions. This paper adopted a macro and micro bibliographic method, which is used to map out the general research landscape. This is followed by a more in-depth analysis of the fields of sustainable design, sustainable construction, sustainable development, and life cycle assessment (LCA). The results show that the combination of AI and BIM helps to make optimal decisions on materials, cost, energy, construction scheduling, and monitoring and promotes the development of sustainable buildings in both technical and human aspects so to achieve Sustainable Development Goals 7 (ensuring access to affordable, reliable, and sustainable modern energy for all), 9 (building resilient infrastructure, promote inclusive and sustainable industries, and foster innovation), 11 (building inclusive, safe, risk-resilient, and sustainable cities and human settlements), and 12 (ensuring sustainable consumption and production patterns). In addition, the combination of AI, BIM, and LCA technologies offers great potential to improve building performance, and the future development of AI and BIM integration should not only consider the sustainability of buildings but also consider the human-centered design concept and the health, safety, and comfort of stakeholders as one of the goals to realize the multidimensional development of smart city based on city information model.

1. Introduction

In the architecture, engineering, and construction (AEC) industry, building in-formation modeling (BIM) is widely used for coordination, collaboration, and managing building information [1] and brings new opportunities for AEC [2]. It is defined as an interactive set of policies, processes, and technologies that produce an approach to managing basic building design and project data in digital format throughout the life cycle of a building [3]. In addition, the term building information management is sometimes used as a synonym for building information modeling to emphasize the importance of defining and managing information throughout the life cycle, not only the modeling activity [4], as it focuses more on the elements of people, processes, and technology in the diffusion of innovation [5]. The widespread use of BIM is due to the benefits it brings to the AEC industry, such as significantly reducing project costs, time, errors, omissions, rework, safety risks, maintaining duplicate operations, and increasing construction productivity [6,7]. Along with the continuous development of BIM, attention needs to be paid to the sustainability aspect of BIM. Due to the rising energy costs and environmental impact, there is a growing drive to develop sustainable buildings [8,9,10]. The theme of sustainable building is to design, construct, operate, maintain, and dismantle buildings while consuming fewer resources and without destroying ecosystems and disrupting the natural rhythm of life [11,12]. Moreover, sustainable AI is already emerging in the field of sustainability, and on the technical level, emerging artificial intelligence (AI) technology has been deemed a complement to BIM [13].

The concept of AI was first proposed at Dartmouth College in 1956 [14]. It is generally interpreted as the ability of machines to learn from experience, adapt to new inputs, and perform human-like tasks [15]. Since the 1970s, AI has expanded into re-search areas, including mechanical theorem proving, machine translation, expert systems, game theory, pattern recognition, machine learning (ML), robotics, and intelligent control [16]. With the rapid development of the internet, the development environment of AI has been changed, and AI has been promoted to the era of AI 2.0 [17]. In addition, artificial intelligence represents a groundbreaking technology that profoundly and positively transforms architectural practice, which facilitates enhanced education, improved health outcomes, and stronger communities, ultimately contributing to the creation of a healthy and sustainable future for our world [18]. At present, AI is being rapidly applied in the AEC field [19], and it has brought considerable benefits to the AEC ecosystem, which is expected to drive important transformations in the industries [20]. Further, AI has been widely used for optimizing building emissions [21], building energy [22], and building structure [23], but it is rarely integrated with BIM [24]. Hence, the application of AI-aided BIM in AEC is still in its infancy [25].

Eleftheriadis et al. [26] recognized BIM’s ability in the field of AEC in their re-search and proposed to embrace AI. Hussein et al. [27] also affirmed the potential of AI and BIM integration, such as improving the efficiency of information queries [28,29,30] and making more effective design decisions [31,32]. Moreover, in line with the Sustainable Development Goals (SDGs), construction is one of the key industries that can build resilient and sustainable infrastructure for human settlements. Recently, SDGs have been shown positive, clear impacts on AI and BIM for sustainable construction and development, such as SDG 4 (ensuring inclusive and equitable quality education and promoting lifelong learning opportunities for all), SDG 7 (ensuring access to affordable, reliable, and sustainable modern energy for all), SDG 8 (promoting sustained, inclusive, and sustainable economic growth, full and productive employment and decent work for all), SDG 9 (building resilient infrastructure, promote inclusive and sustainable industries, and foster innovation), SDG 11 (building inclusive, safe, risk-resilient, and sustainable cities and human settlements), and SDG 12 (ensuring sustainable consumption and production patterns). Thus, by optimizing the management of resources, sustainable development can be achieved, and citizens can be provided with a better quality of healthy life to further realize the construction of smart cities based on the use of manpower, collectivity, and technological capital to promote urban development and prosperity [33]. Moreover, designers are used to carrying out complex designs within limited time and strict rules [34]; as such, appropriate tools to help them make effective decisions are necessary [35,36]. Current studies suggest that the integration of AI and BIM is conducive to the development of the AEC field, but few studies explore the relationship between AI and BIM from the perspective of sustainable building and establish the current research status and future trends in sustainable building, which leaves a gap between AI and BIM technology adoption and development of sustainable building in emerging smart cities and a research question regarding what the relationship between AI and BIM technologies for sustainable building is in the context of smart cities. Therefore, the aim of this paper is to explore the relationship between AI and BIM technologies for sustainable building in smart cities, for which the subsequent two research questions need to be answered, namely, (1) What are the development sequence and current research status of AI–BIM-sustainable building-theme-related research in smart cities? and (2) What are the expected contribution and development trends toward SDGs from the themes of AI–BIM-sustainable building in smart cities? As such, two following research objectives need to be addressed: (1) to explore the development sequence and current research status of AI–BIM-sustainable building-related research in smart cities and (2) to identify the contribution and development trend of the AI–BIM-sustainable building theme to the SDGs in the context of smart cities.

2. Materials and Methods

This paper adopts the bibliometric method to establish the research status and future development trends in the application of AI and BIM technology for sustainable buildings in the context of a smart city. Bibliometric analysis is a method that is used to quantitatively describe and analyze the latest studies and provide researchers with a comprehensive understanding of their research field [37]. In addition, compared with traditional review methods, the adoption of the bibliometrics method can reduce the possibility of subjective judgment by researchers [38], since the analyses are founded on data retrieved from established databases, and not influenced by operators during the analysis process [39]. As such, the bibliometric method has been widely applied in the fields of marketing management [40], biomedicine [41], and physics [42]. However, it has not yet appeared in the research that explores the relationship between the application of AI and BIM technology, and sustainable building and construction.

A bibliometric analysis software platform Citespace5.8.R1, has been selected for this study. The CiteSpace software can analyze and visualize studies in the field of scientific knowledge [43]. It can read bibliographic database resources in different formats, and process document data through time slice, threshold segmentation, modeling, pruning, merging and mapping [44]. After completing these steps, multiple types of bibliometric networks can be constructed, allowing analysts to perform clustering and burst detection [44]. Compared with other bibliometric software, Citespace can objectively generate cluster names by itself, and carry out keyword burst analysis. This paper collects articles from the Web of Science (WOS) core collection database for bibliometric analysis, as the database contains the most important and influential journals in the world, as well as most publications on BIM [45].

The research method flow for this study is shown in Figure 1, which has components of macro quantitative analysis and micro qualitative analysis. The research direction of the relationship between AI and BIM, and sustainable building and construction is established in the macro quantitative analysis, and further explored in the qualitative analysis at the micro level. The process has five steps: (1) Search the core collection database of WOS by combining BIM, AI and sustainability in pairs to obtain a large literature library of union, with search keyword query AK = (((BIM OR “build* information model*”) AND (AI OR “artificial intelligence”)) OR ((BIM OR “build* information model*”) AND sustainabl*) OR ((AI OR “artificial intelligence”) AND sustainabl*)) before year 2024. A large number of documents were obtained with duplicates or little relevance or irrelevant to the research questions, which need to be treated to filter duplicate documents, exclude publications that are not relevant to the research topic of sustainable building, and non-English texts, adopt documents from high-quality databases, preference widely cited articles, and adopt peer-reviewed journal or conference papers, to minimize data bias. A total of 1018 articles were retrieved since year the 1998, with the first article published on AI and BIM in sustainable building, for macro-quantitative bibliometric analysis, of which the most relevant 143 articles were identified for the follow-up micro-qualitative analysis; (2) Based on the results of Step 1, explore the relationship between AI, BIM and sustainable design in the context of the sustainable building; (3) Based on the results of Step 1, explore the relationship between AI, BIM and sustainable construction in the context of sustainable building; (4) Based on the results of Step 1, explore the relationship between AI, BIM and sustainable goals in the context of sustainable building; and (5) Based on the results of Step 1, explore the relationship between AI, BIM and life-cycle assessment (LCA) in the context of sustainable building. Among the five steps, the first step is macro quantitative analysis, and the second to fifth steps are micro qualitative analysis, for which the qualitative content analysis method has been used for the key themes of AI-BIM-sustainable building in smart cities on three qualitative parts, namely, strategy, approach, and method, based on the results of the macro quantitative analysis.

3. Results

3.1. Bibliometric Method with Macro Quantitative Analysis

3.1.1. Trend of Publications

As shown in Figure 2, the first article on AI and BIM in sustainable building was published in 1998. The subsequent literature publications have increased slowly from 8 articles to 22 articles every 10 years (from 2008 to 2017). However, the number of articles per year has increased 23 times in the past seven years from 14 articles in 2018 to 328 articles in 2024. In recent two years, studies have dramatically increased by about 1.6 times from 197 in 2023 to 328 in 2024. Hence, BIM, AI and sustainable building research are currently hot topics.

In addition, the top 10 most productive sources for AI and BIM in sustainable building are shown in Figure 3. Among the sources, Sustainability is the most productive journal, with 217 articles accounting for 21% of all the publications, followed by Buildings (44 articles (4%)), The Journal of Cleaner Production (26 articles (3%)), IEEE Access (23 articles (2%)), Sustainable Cities and Society (21 articles (2%)), Automation in Construction (16 articles (2%)), Energies (16 articles (2%)), Applied Sciences Basel (15 articles (1%)), Environment Development and Sustainability (12 articles (1%)), and Sustainable Development (12 articles (1%)). To be part of the top 10 most productive sources, they require at least five articles to have been published. The top 10 most productive sources play a key role in establishing the foundation for research regarding AI and BIM in sustainable building.

3.1.2. Keywords Co-Occurrence Analysis

Keyword co-occurrence analysis uses keywords in articles to study the conceptual structure of a field and helps to identify relationships between different fields [46]. Keywords are the themes and focuses of the research, and the hot spots and evolution trends of the research field can be obtained through keyword analysis [47]. The down-loaded articles were imported into Citespace software, and the time-to-time slice was adjusted from January 1998 to 2024. This analysis was conducted in a one-year slice. In the process of keyword co-occurrence, synonyms are combined, such as BIM = building information modeling and BIM = building information model. Nodes represent key-words, and the larger the nodes, the higher the frequency of keyword occurrence. The color of nodes changes from orange to purple, indicating the time of co-occurrence from far to near [47]. As shown in Figure 4, AI, BIM, design, construction, management, system, framework, and optimization are the keywords that appear more frequently, and other keywords, such as model, sustainable design, sustainable construction, sustainable building, sustainable development, LCA, IoT, big data, smart city, circular economy, performance, decision making, and decision support system, also have high frequency, which are the research hotspots on AI and BIM in sustainable building. Considering the data in

Table 1. AI and BIM appear most frequently in Citespace analysis keywords of sustainable buildings in the WoS core collection database (devised by the authors).

Frequency	Centrality	Start Year	Keywords
537	0.13	1998	artificial intelligence
207	0.12	2008	sustainable development
88	0.01	2019	machine learning
86	0.05	2009	design
86	0.03	2014	system
84	0.04	2017	management
81	0.07	2013	model
72	0.01	2018	artificial intelligence (ai)
70	0.02	2019	sustainable development goal
66	0.04	2017	performance
66	0.13	2010	building information modeling
62	0.04	2018	big data

, AI (frequency 537) and sustainable development (frequency 207) have the top frequency, and AI (centrality 0.13), BIM (centrality 0.13), and sustainable development (centrality 0.12) have the top centrality, indicating that most of the keywords are related to AI, BIM, and sustainable development. However, the centrality of AI is the same as that of BIM, but the frequency is in the top 537, which is consistent with the fact that AI has more connections with other keywords, as shown in Figure 4. Additionally, except AI and BIM, sustainable development (frequency 207), machine learning (frequency 88), design (frequency 86), system (frequency 86), management (frequency 84), model (frequency 81), sustainable development goal (frequency 70), performance (frequency 66), and big data (frequency 662) are also relatively concentrated research hotspots in the current research field, as shown in Table 1. Management and performance appeared in 2017, indicating that in the process of building development, researchers gradually began to recognize the importance of management and building performance issues for sustainable building, which also ex-plains the reason for the increase in the number of published articles in 2017, as shown in the center of Figure 4. Machine learning, design, model, system, and big data are approaches and technologies to aid AI and BIM in sustainable building toward sustainable development. The terms big data and machine learning first appeared in 2018 and 2019, respectively, with centralities of 0.04 and 0.01, indicating that these cutting-edge technologies have rapidly become topical technologies in the field, which also proves that the growth of the literature quantity after 2018 could be driven by new technologies to achieve sustainable development goal that appeared in 2019 with centrality of 0.02, as shown in Figure 4.

3.1.3. Keywords Cluster Analysis

In order to find the relationship between high-frequency keywords, keywords are clustered, and the Log−Likelihood Ratio (LLR) algorithm is used to extract cluster names. Keywords cluster analysis is used to detect and analyze the emergence and mutation of research trends over a period of time, and to identify the focus of a research trend within a specific period of time [45]. The Modularity Q value is 0.5586; greater than 0.3 indicates that the clustering structure is significant. The Mean Silhouette value of 0.8511 is greater than 0.7, which suggests that the clustering is an efficient, credible and high-quality clustering according to the studies of Liu et al. [48] and Deng et al. [49].

As shown in Figure 5, among the nine clusters of keywords, cluster 1 (emerging technologies), cluster 3 (BIM) and cluster 5 (data-driven technologies) are associated with Information Technologies (ITs) used to achieve sustainable building. Cluster 0 (calculation tool) and cluster 6 (compressive strength) are popular functions of AI and BIM used for sustainable building. Cluster 4 (sustainable development) and cluster 7 (regarding sustainability) are related to sustainable concepts. Cluster 8 (research) shows that the use of AI and BIM for sustainable building is currently research driven, and Cluster 2 (business model) plays a key role in the use.

3.1.4. Keyword Burst Analysis

In CiteSpace, bursting with a high-frequency change rate is generated through burst detection to identify the frontier of discipline research, which has the characteristics of dynamic synthesis [43,50]. The burst detection algorithm, proposed by Klein-berg [51], can identify emergent terms regardless of how many times their host article has been cited. The TOP 13 keyword emergent words for the Strongest Citation Bursts are made as shown in Figure 6. Among them, sustainable design and BIM, including building information modeling and building information model, as shown in Figure 6, are the two keywords with the highest ‘strength’ of emergence, indicating that they have a high research hotspot in the past. Additionally, machine, future, big data, implementation, design, embodied energy, and LCA are the most recent emergent words over the last five years, of which the big data and LCA have the highest burst intensities, and the outbreak period of machine since 2022 still continues, although it has the lowest strength.

3.2. Qualitative Analysis

This section presents the results of a further micro qualitative analysis based on the results above of the macro quantitative analysis. Through searching, screening and summarizing the articles, the relationship between AI and BIM in sustainable design, sustainable construction and sustainable development is further investigated to reveal the development status of AI and BIM in sustainable building and construction, followed by a content analysis in LCA to show the current development hotspots and future development trends in the field.

3.2.1. Artificial Intelligence (AI) and Building Information Modeling (BIM) in Sustainable Design

Sustainable building design is about creating buildings that have minimal impact on the environment by integrating the structure of the building with the ecosystem of the biosphere [52]. It depends on the ability to gain insights into building performance through analysis and optimization of design [10]. The studies on AI technology and BIM for sustainable design begin to emerge gradually after 2014. The case study [53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72] is the main research method used for the most studies. Various AI technologies, such as machine learning (ML) [54,73], Genetic Algorithms (GA) [74], Evolutionary Algorithms (EA) [70], and deep learning [75] are employed to address various optimization problems in architectural design.

The most applications of AI and BIM in sustainable design are for the selection and optimization of building materials, and the exploration of low building costs. The advancement of building energy analysis and machine learning technologies is anticipated to enhance prediction accuracy, optimize energy efficiency, and improve management processes, of which progress holds the potential to facilitate the transition to clean energy and integrate with digital twin technologies, thereby enhancing management practices and reducing energy costs in pursuit of building sustainability goals [76]. An innovative contextual topic modeling approach integrates LDA, BERT, and clustering techniques, enabling the identification of primary scholarly topics, sub-themes, and interdisciplinary themes in scientific research related to sustainable AI in the energy sector [77]. DesignBuilder simulation is utilized alongside BO-LGBM (Bayesian optimization-LightGBM) and an explainable method based on SHAP (SHapley Additive explanation) to predict and clarify building energy performance. Additionally, the AGE-MOEA algorithm is employed for multi-objective optimization (MOO) in the context of uncertainty sources, which has been validated through a case study focused on green building design [78]. In 2012, Porwal et al. [79] integrated AI with a simulated annealing heuristic algorithm to optimize and analyze the use of reinforcement through extraction and analysis of BIM data, to reduce the reinforcement pruning loss and analyze the optimal length of composite reinforcement. Amidst the development of ITs, GAs have also been applied to sustainable design. By coordinating and integrating with parametric design and modularization, building materials are assigned in the design stage to reduce the waste of building materials [64]. In terms of material optimization, on the basis of BIM and parametric design, Wu et al. [70] further applied EA (i.e., GA) to improve the automation and intelligent planning of floor tiles, and extend material planning to the automation and accurate layout of two-dimensional materials, to promote the sustainability. Additionally, Liu et al. [62] integrated BIM and greedy algorithms to achieve the automation of the wood design and planning process, providing a new AI method for two-dimensional material planning for building materials. In multi-objective optimization (MOO), a non-dominated sorting genetic algorithm-II (NSGA-II) based on non-dominated sorting multi-objective GA [80] is widely used. For example, the optimal material selection and minimum construction cost can be achieved simultaneously based on BIM and NSGA-II [55], as well as the MOO of construction duration and optimal material selection in low-income housing projects [72]. The optimal selection of materials minimizes construction cost and duration, while achieving the largest LEED score in a low-income housing research project in Egypt [53], and measuring the environmental sustainability of the building system [55], to achieve sustainable design. Hence, different types of materials and different building components [71] can be selected and evaluated to achieve the most economical and sustainable design concept [74]. Rooftop photovoltaic (RPV) systems present an effective option for facilitating urban energy transition by making use of underutilized rooftop areas to satisfy decentralized energy requirements, which promotes sustainable development through comprehensive urban energy evaluations and offers direction for local energy planning initiatives [81]. High-rise buildings that are self-sufficient and combine resource-efficient usage with electricity generation alongside high-density living could serve as a sustainable approach to future urban development [82]. However, in many studies, the cost is not only related to material selection and optimization, but also to energy and carbon emissions. For example, Dawood M H. [74] applied BIM and GA to residential design based on life-cycle cost (LCC) to check the relationship between the LCC and energy consumption, and ultimately find the best or closest to the best economic and environmental solution. In addition, the integration of improved particle swarm optimization (PSO) and BIM is also used to solve the tradeoff between the life-cycle cost and life-cycle carbon emission during building design [11]. Further, in complex environments, multi-objective particle swarm optimization (MOPSO) can be implemented with BIM to optimize the design scheme, balancing the conflict between project cost and carbon emissions, and focusing on customer satisfaction and environmental health [57]. The utilization of deep learning, IoT, and immersive experience technologies has the potential to yield innovative advancements in the future integration of BIM, sports, and facilities [83]. Therefore, the integration of various AI technologies and BIM has been employed in the context of sustainable design in a wide range of optimization applications for building materials and costs. In a more far-reaching study, the use of AI and BIM is also invoked to indicate the circular economy strategy in the built environment [84].

Moreover, the implementation of AI and BIM to reduce energy consumption and carbon emissions during the design stage, seems to be a very popular research field. On the energy side, energy simulation predictions are made by analyzing various information exchange scenarios at different Level of Details (LOD) and connecting them to an ML model using BIM data [54]. In addition to energy prediction, Shadram et al. [56] focus on the balance between embodied and operational energy in the design process of BIM driven random population-based algorithms such as GA, and conducts sustainable design strategy by exploring trade-offs. In the context of multi-standard decision-making, the optimization model created by BIM can achieve the optimal level of energy use in design scenes through the application of the decision-tree principle [69]. A three-stage research approach is implemented, including data gathering and performance evaluation, BIM, and machine learning techniques for energy prediction, for which four advanced machine learning algorithms are employed to estimate daily energy output [85]. Creative integration of cloud technology with BIM and NSGA-II can also achieve optimization of building energy efficiency, from which the cloud-based energy analysis tools make the optimization process more rapid and accurate [36]. In practice, based on BIM data, Santiago P. [67] explores the potential of using EA to generate facade sun orientation optimization solutions, to simulate the evolutionary optimization of energy and comfort building facade forms in an urban environment. For the study of carbon emissions, Eleftheriadis et al. [63] propose to integrate BIM with NSGA-II algorithm to evaluate the impact of heuristic structural optimization design on carbon performance in the life-cycle of buildings, of which the building design process includes two different objectives: architectural design and structural design [79]. A future building information system platform has been developed and focused on the circular economy, conducting experiments to compare the enhanced IU-Net model with alternative models in order to assess accuracy and other performance metrics, which identifies the particular information related to future building construction from the perspective of sustainable development [86]. As such, an integrated design method based on BIM could be developed to optimize the architectural design and structural form of the building in order to minimize the embodied and operational carbon of the building throughout its life-cycle, the method assists that in saving about 30% of the carbon emissions from energy use during building operations and 21% of the carbon footprint of building materials in a case study of a T-shaped house [66]. In general, the optimization of energy use to carbon emissions generated is often used to compare the environmental impact of various construction projects [87]. Digital technologies are transforming buildings into more integrated, adaptable, energy-efficient, intelligent, and sustainable structures by optimizing resource use, improving operational efficiency, and reducing environmental impact [88]. Therefore, efficient energy utilization to minimize the environmental impact towards sustainability has become a main goal of building design [61].

In addition to the wide application of material costs and energy carbon emissions, construction techniques, staffing and other elements need to be considered in the planning of sustainable building projects [58]. Sophisticated communication networks and IoT sensor technologies are crucial for improving energy efficiency through the monitoring and management of these ecosystems, for which a reinforcement learning (RL) strategy has been introduced aiming at optimizing energy usage in multi-functional buildings within the Energy Plus simulation framework [89]. Thus, more extensive research has been conducted in relation to MOO. Yan et al. [61] divide design objectives into quantifiable and non-quantifiable objectives, and develop a architecture MOO (ArchMOO) framework based on parametric BIM and EA with quantifiable and non-quantifiable objectives, which aids in searching Pareto optimization design solutions with multiple non-quantifiable objective metrics to assist designers in their decision-making. Apart from the ArchMOO framework, Simulation of Environmental Impact of Construction (SimulEICon) [59] was also developed based on MOO, which uses Revit architecture and Microsoft Access to integrate with NSGA-II for optimization to find the best alternative for all components considering time, cost and environmental impact [58]. Moreover, a MOO visual programming package based on BIM and NSGA-II called Optimo [60] was developed to build multidisciplinary optimization of performance during the design process, which has been applied in some university courses and research. Aside from the application of multi-objective GA, Antucheviciene et al. [90] integrate BIM with fuzzy logic to deal with uncertainty in decision making in reviewing sustainable design. A proposed method streamlines and standardizes case-based reasoning (CBR) processes while offering support for addressing semi-structured decision-making challenges related to green retrofits, thus fostering sustainable development and intelligent management within the construction sector [91]. Big data analytics has been employed to develop a conceptual framework with IoT for urban underground engineering, which links concealed features with diverse high-level sensing sources and effective predictive model characterization, aiming to reduce construction costs, enhance infrastructure management efficiency, improve disaster preparedness, and provide advanced smart services for communities [92].

Furthermore, the main objective of sustainable building is to minimize the adverse impact of buildings on human health and the environment through pollution/waste reduction and efficient use of resources, while sustainable design takes into account the impact of building operations on the health of residents [74]. The implementation of sustainable development strategies is an important approach to increasing awareness of environmental and health performance [93]. Wang et al. [87] point out the research direction of sustainable buildings within the context of a healthy environment and building energy efficiency. In addition, the analysis of coordinated efforts has been underscored to enhance Zero Energy Building (ZEB) efficiency and improve photovoltaic performance, which assists to in setting a framework for smart cities that integrate ZEB with transportation and cutting-edge technologies such as Information and Communication Technology (ICT) and sensors [94]. A pluralist methodology was employed to assess and compare the economic evaluations based on land cover for sustainable, moderately sustainable, and unsustainable urban areas, of which future changes in land cover driven by AI were analyzed using geographic simulation software integrated with geographic information systems [95]. Further, Kriegel and Nies [96] indicate that BIM can help in sustainable design in the aspects of building orientation, volume, day-lighting analysis, water collection, energy modeling, and site and logistics management. However, as shown in

Table 2. Research on the application of AI technology and BIM in sustainable design (compiled from the literature).

Year	Author	Method	AI technology Aided BIM	Aim
2024	Mehraban et al. [73]	Case study	Machine Learning (ML)	Energy performance
2024	Natarajan et al. [75]	Energy prediction	Deep learning via IoT/LSTM	Energy consumption
2024	Di Giovanni et al. [85]	Case study	ML	BAPV systems
2024	Zhang et al. [81]	Case study	Deep learning	Evaluate RPV potential
2024	Bereketeab et al. [89]	Simulation environment	Reinforcement learning (RL)	Importance of advanced communication systems and IoT sensors
2024	Ni et al. [97]	Case study	Deep Learning and GIS	Gauge solar energy potential
2024	Asif et al. [88]	Analyze digital technologies	Digital technologies	Immense potential
2024	Liu et al. [91]	Text information analysis	Case-Based Reasoning (CBR) and Random Forest (RF)	Green retrofitting
2023	Chen et al. [86]	Model	AI/IU-Net model	New green building
2023	Shen and Pan [78]	Case study	Ensemble Learning	Building energy performance
2023	Liu et al. [83]	Bibliometric analysis	Deep learning	Sustainable development
2023	Jin and Bae [94]	Data analysis	ML	Contemporary ZEB research contours
2023	Tahmasebinia et al. [76]	Test and evaluation	ML	Sustainable energy management
2022	Ekici et al. [82]	Case study	ANN models	Design sustainable habitation alternatives
2022	Saheb and Dehghani [77]	Topic modeling	AI with IoT	The pathway toward sustainability
2022	Gupta and Bharat [95]	Pluralist evaluation	AI	Assess the spatial resource and natural capital balance
2022	Liu et al. [92]	Data analytics	AI/IOT	Usage of underground space
2021	Wu et al. [70]	Case study	Evolutionary Algorithms (EA)	Material optimization
2021	Vite and Morbiducci [71]	Case study	Non-dominated sorting genetic algorithm-II (NSGA-II)	Design optimization
2021	Çetin et al. [84]	Review	Artificial intelligence (AI)	Circular economy
2020	Jalaei and Mohammadi [68]	Case study	K nearest neighbor data mining	Green Building Evaluation
2020	Starynina et al. [69]	Case study	decision tree	Energy optimization
2019	Liu et al. [65]	Case study	A hybrid algorithm combining greedy algorithm and particle swarm algorithm	Material optimization
2019	Gan et al. [66]	Case study	Genetic Algorithms (GA)	Minimum carbon emission
2019	Santiago [67]	Case study	EA	Energy optimization
2018	Singh et al. [54]	Case study	ML	Energy forecasting
2018	Marzouk et al. [55]	Case study	GA	Building materials/cost
2018	Shadram and Mukkavaara [56]	Case study	Multi-objective GA	Energy optimization
2018	Liu et al. [62]	Case study	Greedy algorithm	Material optimization
2018	Eleftheriadis et al. [63]	Case study	NSGA-II	Minimum carbon emission/cost
2018	Banihashemi et al. [64]	Case study	GA	Material optimization
2016	Dawood [74]		GA	Minimum carbon emission/cost
2016	Marzouk et al. [72]	Case study	NSGA-II	Schedule optimization/material selection
2015	Liu et al. [11]	Case study	Particle swarm optimization (PSO) algorithm	Minimum carbon emission/cost
2015	Liu [57]	Case study	Multi-objective particle swarm optimization (MOPSO)	Minimum carbon emission/cost
2015	Inyim et al. [58]	Case study	GA	Building environmental impact optimization
2015	Yan et al. [61]	Case study	EA	Design optimization
2015	Zhu et al. [59]	Case study	NSGA-II	Design optimization
2015	Antucheviciene et al. [90]	Review	Fuzzy logic	Design decisions making
2015	Asl et al. [60]	Case study	NSGA-II	Design optimization
2014	Marzouk et al. [53]	Case study	NSGA-II	Building cost
2014	Asl et al. [35]	Case study	NSGA-II	Design optimization
2012	Porwal and Hewage [79]	Case study	Combination of simulated annealing heuristic algorithms	Material optimization

, there are still gaps in the current research on the application of AI technology and BIM for sustainable design regarding the reduction in water demand in buildings, and site and logistics management, since BIM, as a data source, can be integrated with various AI technologies to achieve the goal of sustainable design and ultimately achieve the sustainability of the building. The building information has been integrated with height characteristics to assess inter-building shading and utilized the Convolutional Neural Network to detect rooftop obstacles from high-resolution satellite images, which enhances the accuracy of renewable energy development and contributes to sustainable urban planning approaches [97].

3.2.2. AI and BIM for Sustainable Construction

The majority of the studies were published after 2017. In terms of research methods, case studies [98,99,100,101,102,103,104,105] and review [106,107,108,109] are the main research methods in AI and BIM for sustainable construction.

The integration of BIM and AI reveals new value in managing construction projects that are characterized by complexity and uncertainty [110]. Modular construction and prefabrication construction are popular implementation fields in sustainable construction. The advantages of modular construction in construction time cost, worker safety, cost, and less environmental impact [111] enable modular and prefabricated construction to gradually replacing traditional site construction [112]. The adoption of BIM further illustrates the advantages of modular construction in reducing noise levels, operation energy consumption and energy from a quantitative perspective [113]. As ITs are increasingly deployed in integrated construction, the greatest value of building information management lies in management information [114]. The integration of AI technology will further promote the environmentally friendly advantages of modularity, resulting in achieving sustainable construction. In the BIM automatic generation technology of precast concrete bridge deck panels, the rapid determination of the optimal sensor position and the scanning of the target panel can be executed through a greedy algorithm [115]. The BIM model adopts a building-by-building methodology to precisely evaluate material inventories within the building system by leveraging local databases, which enhances the detail of system composition data with the help of which, employing advanced ML techniques, such as linear regression and neural networks, the model is capable of processing both categorical and non-categorical data [116]. In addition, ML is a useful tool for determining the optimal location of sensors through a framework combining the concept of smart assets and distributed manufacturing methods to reduce the environmental footprint of construction sites to achieve sustainable modular construction [106]. Samarasinghe, et al. [101] implement fuzzy logic and BIM with clustering to determine the optimal number of modules and the point at which modules are divided. In the process of module manufacturing, computer vision technology is also introduced with BIM technology to build a steel frame manufacturing pre-inspection system to correct all possible defects in the frame [100]. A multi-dimensional BIM platform based on Radio Frequency Identification (RFID) equipment has been applied with PSO, ant colony algorithm and GA to achieve planning, scheduling, internal logistics and production, for promoting the production of prefabricated parts [117]. In modular construction, the application of BIM based GA can shorten the process of repetitive linear operation in prefabrication construction, to improve productivity [107]. Wang et al. [103] further propose an Improved Genetic Algorithm (IGA) to carry out assembly sequence planning and optimization of precast concrete by taking full advantage of the parameterization of BIM. However, this method would no longer be efficient when the number of components is large. In terms of the progress of modular construction, Zheng et al. [99] developed a module detection model based on the mask region of convolutional neural network (Mask R-CNN) by integrating virtual prototype technology and transfer learning technology, in which the resulting model has been trained with datasets composed of virtual and real images and applied to two modular construction projects for automatic progress monitoring. Further, the Mask R-CNN can be associated with the DeepSORT advanced vision algorithm to conduct automatic detection and monitoring on prefabricated walls in surveillance video during the construction phase [98]. Moreover, the large language model ChatGPT (Chat Generative Pre-Trained Transformer) was integrated into the construction process to build RoboGPT, which was used for automatic continuous planning of robot assembly.

On a traditional construction site, AI and BIM technology can be used for construction progress and safety monitoring, the main approach of which is to process the data obtained by laser scanner through ML. The real 3D progress information has been compared with 4D BIM planning data to obtain the measurement of construction progress [118], where the 4D BIM supports visual monitoring of on-site communication and construction progress. Additionally, the application of AI represents a suitable strategy for enabling large-scale management of sites with varying levels of complexity [119]. By identifying deviations from the 4D model, the effectiveness of on-site environmental impact monitoring can be improved, thereby accomplishing goals of on-site management of safety, workspace, and waste [120]. In the framework proposed by Rahimian et al. [108], a computer vision algorithm is developed with ML to promote the information flow between the construction site and the BIM model, where a game engine is used for integration in the VR environment that allows users to actively participate in the progress assessment and make possible on-demand schedule, and it is believed that future research can use emergency Unmanned Aerial Vehicle (UAV) technology to capture on-site images and reduce safety and health risks. Tibaut et al. [102] compare the information obtained through computer vision algorithm with 4D BIM planning information by applying the UAV to construction progress monitoring, which assists in achieving real-time progress and construction waste monitoring and sustainable management, and even reduce threats to the public environment and human health [121], and improve the health and safety management of construction projects [122]. In addition, as shown in

Table 3. Research on the application of AI technology and BIM in sustainable construction (compiled from the literature).

Year	Author	Research Method	AI Technology Aided BIM	Implementation
2024	Um-e-Habiba et al. [124]	Comprehensive evaluations	ML/control systems	Balance between complexity and control performance
2024	Hosseini et al. [132]	Case study	Neural networks	Reducing maintenance costs throughout the building’s life cycle
2023	Tang et al. [123]	Case study	Principal component analysis (PCA)	Construction and operation of large stadiums
2023	Turayanond and Prasittisopin [126]	Review projects	ML	Maximize the process efficiency
2022	Pan and Zhang [110]	Bibliometric analysis	AI	Automation and digitalization
2021	Turner et al. [106]	Review	ML	Modular construction
2021	Wang et al. [98]	Case study	Convolutional Neural Network	Modular construction progress monitoring
2020	Luo et al. [107]	Review	GA	Prefabricated structure/prefabricated construction
2020	Rahimian et al. [108].	Review	Computer Vision Algorithms/ML	Construction progress monitoring
2020	Sacks et al. [127]	Concept model	ML	Whole process construction
2020	Zheng and Pan [99]	Case study	Convolutional Neural Network	Modular construction
2019	Tang et al. [109]	Review	ML	Construction progress monitoring
2019	Martinez et al. [100]	Case study	Computer Vision Algorithms	Modular construction
2019	Samarasinghe et al. [101]	Field study/Case study	Fuzzy logic	Modular construction
2018	Tibaut et al. [102]	Experiment/Case study	Computer Vision Algorithms	Construction progress monitoring
2018	Wang et al. [103]	Case study	Improved genetic algorithm	Prefabrication/Assembly Order
2017	Li et al. [117]	Pilot studies	PSO/ant colony algorithm/GA	Construction progress monitoring
2015	Liu et al. [104]	Case study	PSO	Construction scheduling
2015	Moon et al. [105]	Case study	GA/Fuzzy logic	Construction scheduling
2013	Kim et al. [119]	Model development	ML	Construction progress monitoring

, the possibility of applying deep learning to object recognition and tracking in the future has been shown in a number of cases. The collected indicators and data have been assessed using regression analysis and principal component analysis to elucidate the influence and significance of each indicator within the evaluation framework [123]. Tang et al. [109] propose to use big data and ML technology to analyze and process BIM and the internet of things (IoTs) data to aid intelligent monitoring and driving. In terms of construction scheduling, Liu et al. [104] suggest a comprehensive scheduling method based on BIM, after which an optimal activity level scheduling of construction projects is automatically generated through PSO algorithm, and an artificial neural network will be further integrated into the system in future work. In the specific scheduling optimization, a fuzzy algorithm and risk simulation model have been developed to explore the automatic search of construction schedule overlap risk assessment, in which the GA is used to optimize the schedule overlap risk after the assessment of job overlap risk [105]. The big data, data acquisition, automation in the construction industry, digital energy management, and building energy modeling are varied and, involve control systems that prioritize occupant needs, ensure energy security, offer flexibility and reliability, and incorporate ML for enhanced control mechanisms [124]. Recently, natural language processing technologies, such as ChatGPT, have been considered to have great potential for completing preliminary and time-consuming tasks in construction project scheduling. The above integrated optimization model ensures the sustainability of scheduling, and makes construction sustainable, as sustainable scheduling refers to the continuous optimal allocation of time and resources throughout the life-cycle of a project [125]. Interestingly, significant gaps in Design for Manufacture and Assembly (DfMA) related to Design for Fabrication (Dfab) and Design for Additive Manufacturing (DfAM) could be addressed by incorporating aspects such as product structural performance, management strategies, case studies, BIM, and ML, which enhance operational efficiency and promote sustainable practices [126]. Additionally, the adoption of Dfaband DfAM technologies positively impacts construction practices in three key areas: economic; social; and environmental, which contribute to a reduction in labor costs that is beneficial for addressing some of the most pressing global economic challenges [126]. In the application of BIM, digital twin (DT) usually appears when the virtual model is connected to the target physical part and the corresponding 3D model is made for the target physical part. Thus, in the whole process of construction, a DT construction concept has been established to develop a coherent, comprehensive and feasible planning and control workflow for the design and construction using the DT information system. It uses AI with ML to analyze the four stages of the planning, implementation, inspection and action cycle to achieve the goal of sustainable construction [127]. The DT was proposed in 2003 by Grieves [128]. In 2012, the concept of DTs was defined by NASA as an integrated multi-physical, multi-scale, probabilistic simulation constructed system, which uses the best available physical model and sensor updates to reflect the life-cycle of the twin of the corresponding system [129]; it converts uncertainty into probability to evaluate whether to perform the task [130]. Thus, the application potential of DT in the AEC field is huge [131]. BIM offers comprehensive digital representations of structures, which not only conserves time and resources but also improves the capacity to anticipate and mitigate future problems, ultimately prolonging the lifespan of building elements and enhancing overall performance, which can greatly benefit the construction and facilities management sectors by providing a more dependable and sophisticated approach to maintaining building health and safety, thus promoting sustainable and cost-efficient management practices [132]. In building construction, the application of modular building is one way to reduce the impact on the environment [106]. The monitoring of construction progress is a way to respond to the monitoring of the environment. Construction scheduling is a way to reduce construction time and optimize resource allocation to achieve sustainable construction. Moreover, the application of the emerging AI technology represented by ChatGPT in sustainable construction also needs to attract the attention of the AEC industry. Therefore, the application of AI and BIM in construction can facilitate achieving sustainable construction. In addition to AI and BIM, new ITs such as IoTs, VR and DT, will be powerful tools to promote sustainable construction. Additionally, the internet has set off a wave of VR in recent years. The concept of metaverse has been proposed and applied in various fields [133], while the concept of metaverse was reportedly inspired by DT [134], as such metaverse could be a new platform for sustainable construction in the development of AI and BIM with the above-mentioned technologies.

3.2.3. Application of AI and BIM to Implement Sustainable Development Goals (SDGs) in the Construction Industry

The implementation measures of sustainable development have been established in the United Nations 2030 Agenda for Sustainable Development [135], which sets out 17 Sustainable Development Goals (SDGs), supplemented by 169 targets and 231 global indicators. The 2030 Agenda and its SDGs offer new opportunities for the construction industry [136] that have a great potential and responsibility to achieve the 17 SDGs [137]. Hence, research covering the relationship between AI and BIM in the construction industry within a sustainable development context was mapped directly or indirectly to the SDGs and was examined by several studies, including Engberg-Pedersen [138] and Morton et al. [139]. A number of research methods have been employed in the studies on the implementation of SDGs through AI and BIM. These methods comprises case studies [56,88,140,141,142,143,144,145,146,147,148,149,150], simulation studies [151,152], experiment [153,154,155], prototyping [156], reviews [90,125,157], and test and evaluation studies [68,71,158], of which the case study is the most used one. The studies that, associate AI and BIM with sustainable development, are focused on SDGs 7, 9, 11, and 12, which are presented in following sections.

• Application of AI and BIM to implement SDG 7.

The application of AI and BIM has been assisted to implement SDG 7 (ensuring access to affordable, reliable and sustainable modern energy for all), as shown in

Table 4. Application of AI and BIM to implement SDG 7 (compiled from literature).

Year	Author	Method	AI Technology Aided BIM	Aim
2024	Li et al. [162]	Case study	AI	Net-Zero Carbon City
2024	Xiong et al. [159]	Computational simulation	AI	Urban block design performance driven optimization
2024	Asif et al. [88]	Case study	ML, AI, Digital Twin	Digital technology in sustainable building
2023	Jin and Bae [94]	Artificial intelligence model	ML/AI	Zero energy building
2023	Accardo et al. [161]	Case study	ML	Comfort conditions and sustainability
2022	Li et al. [152]	Simulation study	GA-NN	Building energy forecast
2020	Fokaides et al. [160]	Review	AI/ML	Sustainable built environment
2020	Dahmane et al. [155]	Experiment	NSGA-II	Comfort with optimized energy consumption
2020	Mataloto et al. [150]	Case study	Long and short term memory neural network	Forecast heating and cooling systems
2019	Garcia and Kamsu-Foguem [147]	Case study	FP-growth algorithms	Assess the usability of buildings and the built environment
2018	Shadram and Mukkavaara [56]	Case study	Random population algorithm	Optimization of energy use
2018	Rinaldi et al. [145]	Case study	AI	IoTs energy efficiency
2017	Salimzadeh and Hammad [143]	Case study	Multi-objective GA	Optimize the location of photovoltaic panels
2015	Eleftheriadis et al. [141]	Case study	Custom genetic algorithm	Structural optimization of steel structure design

. Eleftheriadis et al. [141] developed an overall structural optimization solution framework based on BIM, structural analysis, LCA, and custom GA to seek efficient and environmentally friendly steel design structure scheme and reflect the optimization of energy efficiency from the analysis of structural efficiency and environmental impact. By considering climate change mitigation, Shadram et al. [56] use MOO-based BIM and random population algorithm to achieve the balance between optimized embodied energy and operational energy in the process of building design and reduce total energy use. In addition, energy prediction optimization is an effective way to reduce energy use. Based on parameterized BIM, an automatic simulation of the energy database can be carried out, and nonlinear fitting is conducted by a genetic algorithm-neural network (GA-NN) model to predict building energy consumption [152]. By optimizing the use of resources, improving operational efficiency and minimizing environmental impact, digital technologies make buildings more integrated, flexible, energy-efficient, intelligent and sustainable. Digital technologies have been predicted to reduce energy intensity in the building industry by 30% to 50% over the next 20 years [88]. In terms of the research on energy efficiency, Rinaldi et al. [145] integrate BIM with IoTs to improve the optimization of energy efficiency through self-learning and analysis of data by AI and cognitive computing. Performance-driven optimization (PDO) has been used to synthesize the various quantifiable properties of buildings to create sustainable, resilient, and adaptable urban spaces that align with sustainability principles and ensure that they are both aesthetically pleasing and functionally efficient [159]. Fokaides et al. [160] further apply DT on the basis of the IoTs to improve the energy efficiency of buildings through AI algorithm and ML, and believe that future research would be devoted to the transformation of buildings into intelligent units to smoothly integrate them into smart cities. Existing studies also take human health and comfort into account when focusing on energy efficiency. Considering expert opinion based on residents’ health and comfort satisfaction, frequent pattern mining using the FP-growth algorithm can help to identify unobserved energy consumption patterns related to user behavior or weather constraints for future modern energy regeneration consideration [147]. In the process, data from IoT devices has been linked to the BIM model of the test case building using an existing asset management platform that integrates BIM and geographic information systems (GIS) to create a digital shadow that collects real-time data on indoor conditions, including temperature, humidity, and carbon dioxide levels [161]. Since the tracking progress of the NZCC (Net-Zero Carbon City) requires advanced tools, such as city DT and GIS based spatial analysis, AI, and big data, for collection and analysis to predict and monitor, GIS and BIM can be utilized to assess concealed modern energy for carbon emissions and forecast urban development-related emissions [162]. Zero energy building concept as a sustainable solution to high energy consumption, public research and development funding data have been used to reveal the continuous development process [94], while, Dahmane et al. [155] used BIM database to automatically deploy optimal sensor network framework for intelligent buildings. The framework constructs an EA (NSGA-II) to solve the optimal deployment of the number of sensors in the sensor network, and the collected data are used to optimize resource consumption and improve indoor residents comfort. At the same time, the combination of short and long-term memory neural network technology and IoTs technology has also been used in energy efficiency research. The results of IoT sensor measurements are executed and represented in a BIM Model, where real-time user interaction and automated manipulation can predict relevant heating/cooling systems, and sensors can influence user behavior, reduce energy waste and indirectly increase collective comfort, while future research takes the collective feeling as the research’s aim [150]. Building HAVC system is an important user of energy, and its operation and maintenance affect the health and safety of residents [163]. Further, in the research of clean energy, Salimzadeh et al. [143] reconstructed a detailed building surface model with LiDAR and BIM, and optimized photovoltaic layout through multi-objective genetic algorithm (MOGA) to maximize photovoltaic panel installation capacity.

2.

Application of AI and BIM to implement SDG 9;

Regarding SDG 9 (build resilient infrastructure, promote inclusive and sustainable industries, and foster innovation), studies in this field mainly focus on the construction of transportation infrastructure; sustainable industrial development is achieved through innovation in construction methods and techniques. In terms of transportation, Zhang et al. [164] have developed an operation management platform based on BIM technology as a supplement to equipment management. AI also has been applied in metro projects to solve problems such as fleet monitoring and asset maintenance.

3.

Application of AI and BIM to implement SDG 11;

The application of AI and BIM has been facilitated by implementing SDG 11 (building inclusive, safe, risk-resilient and sustainable cities and human settlements), as shown in

Table 5. Application of AI and BIM to implement SDG 11 (compiled from the literature).

Year	Author	Method	AI Technology Aided BIM	Aim
2024	Tamang [168]	Model training	ML/DL	Building inventories
2024	Zhou and Weng [165]	Data engine	ML	Global city map
2024	Taheri and Sobanjo [167]	Model study	ML	Improved monitoring of infrastructure
2023	Wu and Maalek [166]	Model study	ML	Intelligent management of old buildings
2022	Ekici et al. [82]	Case study	Deep learning/ML	Self-sufficiency in energy consumption
2022	Sampaio et al. [169]	Case study	AI	Digital transition in building management
2019	Liu et al. [48]	Case study	Greed algorithms/PSO	Design materials and waste
2016	Marzouk and Abubakr [142]	Case study	NSGA-II	Construction time cost optimization
2015	Wu and Kaushik [156]	Prototype	Pathfinding algorithm	Aging design
2014	Li et al. [151]	Simulation study	Taboos search	The fire location
2014	Wang et al. [158]	Test and Evaluation	Pathfinding algorithm	Fire evacuation
2012	Porwal and Hewage [79]	Case study	Simulated annealing heuristic algorithm	Optimum length of composite reinforcement

. A tailor-made data engine has been introduced to generate high-quality and diverse training samples on a global scale, for which the published global data product and OpenStreetMap data have been used to build a global city map in support of SDG 11 [165]. Marzouk et al. [72] optimize the cost of construction time and materials by implementing BIM with computer simulation and using NSGA-II for affordable home for more people to settle down. The BIM-based framework is assisted with intelligent decisions for aging buildings to assess the environmental impact of aging building restoration efforts to improve overall energy performance [166]. Through CIM tools, advanced technologies in the sustainable design and construction of transportation infrastructure have been integrated and utilized to continuously monitor, analyze, and forecast throughout life cycle stages to maintain and improve the structural integrity and long-term performance of the infrastructure [167]. The implementation of Deep Neural Networks (DNN) can address the limitations associated with deep algorithms, as such accurate building inventories and related information are crucial for sustainable urban governance [168]. The COBie (Construction and Operation Building Information Exchange) model for BIM has been utilized to specify the information required for BIM objects to support facilities management (FM) activities and to enhance interoperability between BIM software and information systems [169]. In the case of building health and safety in fire, Li et al. [151] propose a BIM centered environment awareness beacon deployment algorithm for indicating the location of first responders and trapped people in building fire emergency scenes. Interestingly, Wang et al. [158] use BIM as a building information provider to cooperate with game technology and pathfinding algorithm to build an adaptable VR environment that is with the purpose of enhancing fire evacuation plans throughout the life cycle of the building. In addition, the MUZO (Multi-Zone Optimization) methodology has been developed for self-sufficiency in high-rise buildings at the building scale, and at the building and block scale, highlighting the need to integrate the potential of neighboring buildings [82]. However, it is necessary to further explore whether the shortest evacuation route is the safest. Moreover, in the research on environmental sustainability, as in the research on vulnerable groups, Wu et al. [156] utilized the synergistic effect of BIM and game engine to guide the issue of improving design communication in sustainable aging design projects, where the pathfinding algorithm is used for automatic path planning in the game;

4.

Application of AI and BIM to implement SDG 12.

The application of AI and BIM has been associated with implementing SDG 12 (ensuring sustainable consumption and production patterns). When it comes to technological innovation for production patterns, AI [170] and BIM [171] are new technologies in the construction industry. On the innovation of BIM, Krijnen et al. [154] propose the use of ML to derive implicit knowledge from BIM information, thus providing useful insights for smarter decisions in building design and management. With regard to the building design process, solutions generated by interactive and visual clustering genetic algorithm (IVCGA) are integrated into the BIM environment to enhance design information, and allow for the solutions to be viewed in greater detail as BIM models [140]. Moreover, a design process developed by Yan et al. [172] supports game tools in BIM to enhance building design and visualization by integrating pathfinding algorithms into the game theory. Khan et al. combined the data generated by DesignBuilder simulation with the BO-LGBM (Bayesian Optimization-LighTGBM) prediction model of LIME (locally interpretable Model Unknowable interpretation) technology for energy prediction and analysis with multi-objective optimization technology AGE-MOEA, which significantly improves the transparency of machine learning predictions and effectively identifies the best passive and active design solutions, making an important contribution to sustainable construction practices [173]. As for the optimization of building design, the integration of BIM with GA [174] and NAGA-II [61] can effectively optimize the building structural design framework and sustainable building design under multiple objectives scenarios. Further, concerning building decision-making, the integration of BIM and fuzzy logic effectively solves the problem of uncertainty in decision-making to make sustainable decisions [90,113]. Faghihi et al. [153] developed a method for construction engineers and project managers to directly generate construction sequence using geometric information of the project, which retrieves enough information from BIM and automatically deduces the structurally stable construction sequence using GA to reduce the difficulty of learning how to establish a correct construction project and its schedule. A building information system platform for the circular economy has been developed to improve the traditional convolutional neural network U-Net model, and could be used in the analysis and decision of future buildings [86]. In the absence of uncertainty, the integration of GA and BIM has shown the potential to make better decisions on design component schemes [71].

Through the optimization of materials, the waste of materials can be reduced, the cost can be optimized, and the resource efficiency can be improved, thus promoting sustainable economic development. SimulEICon is a software prototype based on BIM technology with NSGA-II, designed to support the time, cost, and environmental impact of research [59]. The NSGA-II can also be used to select optimal materials to minimize the LCC of building material scenarios [55]. In addition, Liu et al. [144] used a greedy best fit algorithm to optimize the design of the multi-panel structure, where on the basis of making full use of the advantages of building information in BIM, multi-panel design optimization is taken as a typical one-dimensional blanking optimization method to improve resource utilization efficiency. Further, in the construction process, BIM and symbiotic organisms search has been associated with optimizing material layout from the perspective of dynamic task scheduling to improve production efficiency, solve the layout problem of the construction site, reduce the cost of material transportation, and effectively simplify the construction process [149]. Mangal et al. [146] and Wu et al. [70] implemented GA with BIM to achieve the optimization of integrated reinforcement and floor tile with reinforced concrete frame structure, and promote sustainability in the AEC industry through automation and accurate layout design of cutting materials, such as glass. Moreover, Porwal et al. [79] suggest an optimization process for steel bar trim loss, which integrates a simulated annealing heuristic algorithm with structural BIM to minimize steel bar waste in the design stage, and the hybrid algorithm integrating greedy and PSO algorithms are also implemented with BIM, which is used to solve the problem of two-dimensional irregular shape plate cutting for optimizing material cutting plan to minimize the waste of material [65]. In terms of urban waste treatment, Aldebei et al. [157] put forward the concept of urban mining as a development of the stock of urban construction materials to estimate and calculate the waste materials through BIM, AI and ML, which can be used in future mining.

Liu et al. [11] have developed a MOO model with the BIM based simulation system and PSO-based optimization system to help designers identify and select the optimal carbon emission and cost trade-off design scheme. AI technology has been employed to thoroughly extract fundamental information from collected images of old buildings, analyze and design a future-oriented building information system platform focused on the circular economy, which integrates the IU-Net (Fully Invertible U-Nets) model, a comprehensive neural network that offers higher prediction accuracy and enhanced functionality for analysis and decision-making in future building projects [86]. In terms of sustainability rating, as shown in Table 6, AI and BIM technology can assist in evaluating the sustainability of buildings. Mahmoud et al. [148] developed a global sustainability rating tool for existing buildings, by proposing sustainability assessment attributes, determining their weights with fuzzy logic, and considering regional differences. As for the existing rating tool LEED, some studies contribute to an automated method to evaluate the potential sustainability of the buildings based on the LEED certification system and provide a framework for calculating LEED points based on BIM at the concept stage, in which Distance Weighted K-Nearest Neighbor (DWKNN) is used to calculate the missing credits [68]. Further, the content of sustainable production is mainly related to construction. For example, in sustainable construction, the integration of GA and BIM is used to optimize the layout of construction tower cranes [142] and solve the scheduling problem in construction [109], to aid sustainable construction on the premise of ensuring the health and safety in terms of production patterns of construction workers [175].

Therefore, the studies on the application of AI and BIM to drive the enactment of the UN 2030 Agenda for Sustainable Development [135] have made reasonable inroads into the implementation of SDG 7, 9, 11 and 12. Emerging digital technologies, such as the IoTs, DT, and VR, are increasingly being used in the SDGs. Currently, only a few studies focus on human health and comfort, technical education and training of constructors, and opinions and information exchange between stakeholders in the construction process. Future research could focus on people centered healthy, efficient, and sustainable buildings in the context of smart cities, integrating various digital technologies.

3.2.4. AI and BIM in Building Life-Cycle Stages

The life-cycle stages can be defined as a pre-project stage, design stage, construction stage, facility management stage and restructuring and waste recovery stage [176]. The majority of the studies on AI and BIM in sustainable development are focused on the design stage. The advantage of integrating a life-cycle perspective with sustainability assessments goes to occur beyond the completion of a project, to improve engineering design and building performance [177]. However, the pre-project stage seems to be a forgotten land for implementing AI and BIM to assist sustainability. The comprehensive inspection of BIM with transportation and facilities is essential to improving the efficiency, sustainability, and intelligence of buildings and infrastructure, from which the three areas can generate synergies to help better plan, construct, and manage building and infrastructure projects [178].

The applications of AI and BIM in the design and construction stages have been analyzed in detail in Section 3.2.1 and Section 3.2.2. Generative design in the built environment revolutionizes traditional approaches and promotes a more effective and immediate response to design methods, thereby promoting enhanced innovation and sustainability in the design practice [179]. The BIM–MCP approach integrates building information simulation techniques into sustainability assessments for emergency medical engineering and smart healthcare facilities [180]. Zaballos et al. [181] propose the concept of a smart campus under the concept of a smart city. New information and communication technologies make it possible to manage the real-time monitoring of university campus health and environmental conditions, such as pollution, noise, and natural or man-made risks and epidemics, and manage public spaces and facilities to achieve sustainable development via building information management [182]. In addition, integrating techniques and concepts gained from goal-driven case studies across different industries allows for the construction of comprehensive frameworks of DT to monitor and digitally manage the built environment through the entire life cycle of a building facility, including design, construction, operations and maintenance [183]. Although DT in buildings requires a high initial investment and expertise, it has great benefits for the building life cycle, and the application of DT in buildings is currently mainly implemented in the design, maintenance stage and final stage of construction [184]. The DT technology enables sustainable building energy management and cost reduction by monitoring, optimizing, and forecasting building energy consumption in real time [76]. A method has been developed to enable data to be visualized and interlinked with the same database of the BIM model, from which project stakeholders are able to permanently link with the BIM model, and access, and update the data in real time [185]. The case study of BIM integrated with the wireless sensor network (WSN) based on the IoTs has been used in the field of environmental monitoring and emotion detection systems, in which the WSN is an important tool for building information management to gain insight into the health and environmental comfort of occupants through ML analysis [186]. BIM, IoTs and ML are also applied to the predictive maintenance of building components to extend the life of components [187] and effectively support the continuous development of established environmental control and monitoring toward a green and sustainable dimension [188]. The research on the integration of BIM and IoT emphasizes building intelligence through BIM applications, enabling occupants to engage more effectively in sustainable design and decision-making processes for buildings [189]. At the restructuring and waste recovery stage, the BIM has been implemented with GA and artificial neural network to quantitatively evaluate the technical choice of a building renovation project [190]. The application of digitization and advanced technologies to predict the generation of construction and demolition waste, waste identification and sorting, and computer vision for waste management improves waste management and enables a circular economy for buildings [191]. At present, there are few research studies looking across the above two stages, i.e., the facility management stage and restructuring and waste recovery stage. The facility management stage is assisted by the IoTs in conducting the monitoring of health and the environment, while the restructuring and waste recovery stage is associated with the choice of transformation technology.

Table 6. Application of AI and BIM to implement SDG 12 (compiled from the literature).

Year	Author	Method	AI Technology Aided BIM	Aim
2024	Khan et al. [173]	Build model	ML	Sustainable green building design
2023	Chen et al. [86]	Improved model	ML	Innovation in resource utilization
2023	Chen et al. [189]	Test and Evaluation	ML	Circular economy
2021	Vite and Morbiducci [71]	Test and Evaluation	GA	Digital decision model
2021	Wu et al. [70]	Case study	GA	Design material optimization
2021	Aldebei and Dombi [157]	Review	AI/ML	Urban waste building materials
2020	Dasović et al. [125]	Review	GA	Sustainable scheduling
2020	Jalaei et al. [68]	Test and Evaluation	KNN	LEED integral calculation
2019	Mahmoud et al. [148]	Case study	Fuzzy topsis	Sustainability assessment tool
2019	Cheng and Chang [149]	Case study	Symbiotic organisms search	Construction material layout optimization
2019	Hammad et al. [113]	Case study	Fuzzy logic	Decision making tool
2018	Hamidavi et al. [174]	Concept framework	GA	Optimized structure design
2018	Marzouk et al. [55]	Case study	NSGA-II	LCC and environmental sustainability
2018	Mangal and Cheng [146]	Case study	GA	Integrated reinforcement optimization
2017	Liu et al. [144]	Case study	Greedy best fit algorithm	Wall panel configuration optimization
2016	Marzouk et al. [72]	Case study	GA	Tower crane selection decision
2015	Liu et al. [104]	Case study	MOO	Total carbon emissions and costs
2015	Antucheviciene et al. [90]	Review	Fuzzy logic	Design decisions
2015	Yan et al. [61]	Case study	NSGA-II	Optimization of design framework
2015	Krijnen and Tamke [154]	Experiment	ML	Further exploration of BIM
2014	Rafiq and Rustell [140]	Case study	GA	Sustainable design options
2014	Faghihi et al. [153]	Experiment	GA	Construction scheduling
2012	Zhu et al. [59]	Case study	NSGA-II	Design decisions making
2011	Yan et al. [172]	Framework development	Pathfinding algorithm	Games and Teaching

Table 6. Application of AI and BIM to implement SDG 12 (compiled from the literature).

Year	Author	Method	AI Technology Aided BIM	Aim
2024	Khan et al. [173]	Build model	ML	Sustainable green building design
2023	Chen et al. [86]	Improved model	ML	Innovation in resource utilization
2023	Chen et al. [189]	Test and Evaluation	ML	Circular economy
2021	Vite and Morbiducci [71]	Test and Evaluation	GA	Digital decision model
2021	Wu et al. [70]	Case study	GA	Design material optimization
2021	Aldebei and Dombi [157]	Review	AI/ML	Urban waste building materials
2020	Dasović et al. [125]	Review	GA	Sustainable scheduling
2020	Jalaei et al. [68]	Test and Evaluation	KNN	LEED integral calculation
2019	Mahmoud et al. [148]	Case study	Fuzzy topsis	Sustainability assessment tool
2019	Cheng and Chang [149]	Case study	Symbiotic organisms search	Construction material layout optimization
2019	Hammad et al. [113]	Case study	Fuzzy logic	Decision making tool
2018	Hamidavi et al. [174]	Concept framework	GA	Optimized structure design
2018	Marzouk et al. [55]	Case study	NSGA-II	LCC and environmental sustainability
2018	Mangal and Cheng [146]	Case study	GA	Integrated reinforcement optimization
2017	Liu et al. [144]	Case study	Greedy best fit algorithm	Wall panel configuration optimization
2016	Marzouk et al. [72]	Case study	GA	Tower crane selection decision
2015	Liu et al. [104]	Case study	MOO	Total carbon emissions and costs
2015	Antucheviciene et al. [90]	Review	Fuzzy logic	Design decisions
2015	Yan et al. [61]	Case study	NSGA-II	Optimization of design framework
2015	Krijnen and Tamke [154]	Experiment	ML	Further exploration of BIM
2014	Rafiq and Rustell [140]	Case study	GA	Sustainable design options
2014	Faghihi et al. [153]	Experiment	GA	Construction scheduling
2012	Zhu et al. [59]	Case study	NSGA-II	Design decisions making
2011	Yan et al. [172]	Framework development	Pathfinding algorithm	Games and Teaching

Moreover, as shown in

Table 7. AI technology and BIM in building life-cycle (compiled from the literature).

Life Stage	LCA	AI Technology Aided BIM	Aim	Method	Year	Author
Design	LCC	MOPSO/EA	Carbon emissions and costs	Case study	2015	Liu [11]
Design	LCC	PSO	Carbon emissions and costs	Case study	2015	Liu [57]
Design	LCC	GA	LCC and energy consumption	Framework development	2016	Dawood [74]
Design	LCC	NSGA-II	Duration and optimal material selection	Case study	2016	Marzouk et al. [72]
Design		GA	LCC optimization	Case study	2018	Marzouk et al. [55]
Design		ML	Energy analysis		2018	Singh et al. [54]
Design		NSGA II	Design selection	Case study	2021	Vite and Morbiducci [71]
Design	LCA	Fuzzy logic	Material selection	Case study	2021	Figueiredo K et al. [201]
Design	LCA	ML/Random forest (RF) regression algorithm	Environmental benchmark	Case study	2021	Martínez-Rocamora A et al. [200]
Design	LCA	AI/ML	Sustainable smart cities	Case study	2022	Jesus et al. [199]
Design		AI	Architectural life cycle design	Proposed model	2022	Scherz et al. [177]
Design	LCC	Digital Twin	DT technology facility management	Case study	2023	Hodavand et al. [184]
Design	LCA	AI, ML	Sustainable and eco-friendly trajectory	Model introduction	2024	Mao et al. [196]
Design		GD	Enhanced innovation and sustainability	Model intro-duction	2024	Chew et al. [179]
Design	LCA	Digital Twin	Building and infrastructure projects	quantitative research technique	2024	Liu et al. [91]
Design		ML	Building information simulation	Case study	2024	Chen and Laokhongthavorn [180]
Construction		GA	Construction scheduling	Case study	2014	Moon et al. [105]
Construction		GA	Construction scheduling	Experimental Validation and Design	2014	Faghihi et al. [153]
Construction		GA	Construction scheduling	Case study	2015	Moon et al. [105]
Construction		ML	Environmental energy analysis/Occupant residents comfort	Case study	2020	Zaballos et al. [181]
Construction		ML	Parts maintenance	Case study	2020	Cheng et al. [187]
Construction	LCA	Digital Twin	Sustainable energy management	Case study	2023	Tahmasebinia et al. [76]
Construction		GA	Inspection of building components	technical combination	2024	Hosseini et al. [132]
Construction	LCA	Digital Twin	Digital management	Establish a framework	2024	Piras et al. [183]
Facility management		ML	Environmental monitoring and energy	Case study	2021	Tagliabue et al. [188]
Facility management	LCA	AI	Link building dynamic data	Develop data management methods	2022	Rodrigues et al. [185]
Facility management	LCC	ML	Sustainable smart cities	Case study	2023	Chen et al. [86]
Restructuring and waste recovery		GA and artificial neural network	Evaluate technical options for a building renovation project	Case study	2014	Asadi et al. [190]
Restructuring and waste recovery		AI/ML	The construction industry uses digital technologies to promote a circular economy	Case study	2023	Rodrigo et al. [191]

, LCA is a powerful tool for calculating the environmental impact of a building throughout its life-cycle stages [192,193]. It quantifies a set of environmental, social and economic performance indicators by accounting for all inputs, outputs and flows within process, product, or system boundaries [194]. By using ISO 14040 and ISO 14044 [195], the LCA can identify the weakest environmental points and highlight the most environmentally friendly solutions [194]. Theoretically, the LCA can be carried out easily in conjunction with BIM [181], establishing dynamic evaluation indicators, enhancing interdisciplinary research and regional considerations, and introducing LCA to address the challenges associated with indicator setting in the current evaluation system [196], in which the BIM is considered as a potential tool that can greatly improve the information flow throughout the life cycle of a building [197], and the integration of BIM and LCA facilitates the analysis of building projects from a sustainable perspective [198]. The traditional, subjective, time-consuming and labor-intensive building inspection methods have been addressed by integrating BIM and neural networks to propose innovative ways to assess the condition of buildings, improving accuracy, efficiency, and predictive power [132]. LCC is considered one of the three technologies to achieve LCA and is used to measure the level of sustainable development [55]. Liu [57] used MOPSO to optimize LCC and life-cycle emissions in buildings, where the building design strategy was optimized to save LCC and improve the sustainability of building schemes by integrating BIM with MOPSO. In another study by Liu et al. [11], the improved PSO algorithm is conducted with BIM to optimize building LCC. GA has also been used to aid LCC. Dawood [74] implemented GA and BIM to develop a residential optimization design framework based on LCC, which has been used to test the LCC and energy consumption cost, for the selection of the optimal scheme with the minimum LCC of the building, while in the low-cost housing that met people’s living needs, BIM and NSGA-II have been employed to determine the minimum LCC through material selection [72]. Further, an AI-driven adaptive neuro-fuzzy inference system (ANFIS) has been utilized to forecast the environmental impacts associated with the life cycle of industrial water treatment, of which AI methods are employed to enhance LCA models that facilitate the establishment of predictive machine learning frameworks, thereby enhancing confidence in decision-making processes [199]. Martínez-Rocamora et al. [200] calculated environmental impact indicators by the use of LCA tools and ML in BIM, in which the indicators are derived from the changes in its construction plan and continuous improvement of the sustainability of the building. They believe that future research should focus on the flexibility of the LCA database and consider the elements contained in the building extensively. However, Figueiredo et al. [201] proposed the concept of life cycle sustainability assessment (LCSA) and integrated the LCSA, BIM, and fuzzy logic into an innovative scheme to determine the selection of the most sustainable material for building projects.

Although the LCA method can effectively improve building performance, and BIM and LCA integration also has great potential, there is still a lack of digital optimization models for efficient design decisions [198]. Therefore, AI has the potential to further BIM-LCA integration.

4. Discussion

This paper tries to reveal the interesting development of AI–BIM-sustainable building research, and identify AI–BIM-sustainable building theme for SDGs in the context of smart cities, which have challenges and limitations.

4.1. The Development of AI–BIM-Sustainable Building Research in Smart Cities

The results of Section 3 show two main interesting developments of AI–BIM-sustainable building research trends in smart cities, namely, AI and BIM in sustainable design, construction, development, and life-cycle stages, and with new ITs.

4.1.1. AI and BIM in Sustainable Design, Construction, Development and Life-Cycle Stages

The results of Section 3 suggest that AI and BIM can promote sustainable building development in three aspects: sustainable design; sustainable construction; and sustainable development. It is also believed that the inclusion of LCA promotes the development of sustainable building in the context of smart cities in the future.

Nowadays, BIM has become a popular tool for sustainable building design, and AI technology has also been widely applied in the improvement of sustainable building design [11]. BIM, as a sustainable technology, is used to create and monitor the digital information of a project [171], while AI has been applied to pre-process data or infer new information from existing data [202] to optimize and make decisions on materials, costs, energy, construction scheduling, and schedule in the construction cycle, promote the development of technology and humanity in many aspects, and improve building performance through LCA. AI algorithms such as GA, fuzzy logic, ML and PSO, and the emergence of new technologies like chatGPT can not only reduce the workload of designers [57] but also help them achieve sustainable design [74]. In addition, good design decisions are very important for the sustainable development of buildings [11]. Sustainable building design is more complex than traditional design because of the need to address the requirements of environmentally sustainable systems [68], take into account the goal of minimizing energy consumption in sustainable buildings, and realize the economic significance of sustainable buildings [74]. The monitoring of the built environment in sustainable construction can also help decision-makers formulate new strategies to reduce environmental impacts [202]. Therefore, sustainable construction is the supplement and extension of sustainable design, and the two complement each other to achieve the goal of sustainable building. In Section 3.2.3, research on AI and BIM in sustainable development focuses on SDGs 4, 7, 8, 9, 11 and 12. While a study by Goubran et al. [203] on SDGs in construction projects indicates that the SDGs 4, 6, 7, 8, 9, 11, 15 and 17 should be considered in the design stage of the project. As such, sustainable design should further take SDGs 6, 15, and 17 into account and explore the opportunities of AI and BIM in water resources, terrestrial ecology, and partnerships to achieve sustainable buildings.

With the integration of BIM and the optimization model of digital city, the concept of city information model (CIM) is also proposed to further improve the overall efficiency of city management [204]. Moreover, sustainable building can be further established from the perspective of the whole life cycle by integrating with the LCA [200]. However, the development of AI methods is extremely complex and may consume a lot of time, energy, and money in the operation process [205]. One study suggest that when computer performance is insufficient, it can be optimized on more powerful online machines [125]. This may be the current limitation affecting the application of AI technology with BIM [206].

4.1.2. AI and BIM in Sustainable Building with New Information Technologies

In the process of combining AI and BIM in Section 3.2.2 and Section 3.2.3, ITs such as the IoTs, 5G, sensors, and VR also emerged as being applied in the process to promote the establishment of sustainable buildings and the development of smart cities. The integration of various technology based IoTs with AI and BIM is a common practice. In terms of data acquisition, wireless sensor technology [207], RFID [208], and UAV technology [209] can aid in collecting building data and integrating information from different environments directly [102] or through processing into BIM [155,207]. AI is used to optimize the layout of sensors [106] and further processing of acquired information [202] to ensure the reliability of intelligent buildings by collecting the integrity and relevance of information [155]. Additionally, IoT solutions are based on a large number of sensor data [109,209], and the application of 5G technology speeds up the data transmission [210], where IoTs and big data complement each other and provide rich data sources for big data analysis [211]. In terms of data analysis, AI can analyze system defects and determine the ability to build resource requirements [84]. Further, the development of BIM integration with simulation, IoTs and AI processes leads to DT processes [212]. The DT is an emerging technology gaining momentum in the construction industry, which uses advanced IoTs to connect different objects [213] and has limitations to data sharing and visualization [214].

Interestingly, immersive technologies (ImTs) are also widely used in sustainable building. BIM is a semantically rich data platform that can provide data for VR and mixed reality (MR) environments [215]. The integration of emerging technologies, such as point cloud and digital twins (DT), with life cycle assessment, represents a significant area of interest, providing opportunities for innovative breakthroughs in the future [178]. The application of ImTs in the AEC industry establishes an innovative learning style [212] and user-friendly learning environment for AEC professionals [215] and also assists the communication of designers [216]. With the application of AI, the automation of construction progress monitoring can be conducted [108]. The further integration of game engines [158,217] and cloud technologies will also encourage project participants to achieve sustainability goals [216]. Therefore, the integration of ImTs with AI and BIM has great application potential in the whole life cycle of buildings, which is in line with the research conclusion of Alizadehsalehi et al. [218]. When futurists and technologists are exploring the role of the metaverse in different fields, collaborative work is an important application direction [219]. ImTs, gamification, AI and BIM can assist in presenting a better social communication space for designers and users and create value for sustainable building in the metaverse [220], while in virtual spaces and metaverse, AI can be used to develop tasks quickly and continuously for long periods of time [221]. Thus, it provides a new direction for sustainable building development in the post-epidemic era.

The integration of more new technologies in AI and BIM will undoubtedly promote the development of sustainable building. Digital technologies are operable and transformative in efforts to achieve sustainable development on SDGs [222]. However, while considering the advantages of digital technology for sustainable building, it should not ignore the problems of information compatibility, software interoperability and technology costs related to the integration of various new technologies.

4.2. The Relationship Between AI and BIM for Sustainable Building in the Context of Smart Cities in Terms of Stakeholders

The results of micro-analysis in Section 3.2 for revealing the AI–BIM-sustainable building theme to the SDGs in the context of smart cities suggest that the current sustainable building is mainly focused on the optimization of energy, materials and scheduling by AI-BIM technology, and there are few studies on stakeholders. However, as people pay more and more attention to climate change and the environment, residents demand more and more sustainable design of buildings [222]. The BIM provides a common single data source for stakeholders [223], enables information sharing [224], and can also be used to improve workers’ safety and health management [225,226]. AI also enables sustainable design at the design stage [74]. The safety and health issues of stakeholders have been discussed in studies of VR [158] and IoTs [109,181]. The ImTs also have the potential to establish user-centered building information management [227]. The construction safety of construction workers is undergoing severe tests [228], while AI has great potential in safety project management [229], such as the safety layout and management of construction tower cranes [230]. The operation and maintenance of building heating, ventilation and air conditioning systems (HAVC) [163] and building pollutants [121,231] are also research directions related to safety and health of stakeholders. Therefore, human-centered design principles need to be applied to sustainable buildings [232], creating an experience-based dialogue between designers and residents [156]. However, the most construction workers are not well educated and do not know the power of advanced IT technology; as a result, the effect of the implementation of platforms, such as the RFID-enabled real-time BIM platform, is often ignored [117]. Therefore, if AI and BIM are designated to serve stakeholders successfully, it is necessary to give priority to employee education and training [204,233], or to develop BIM-based tools as simple as possible for designers without LCA knowledge [234]. In terms of smart cities, the CIM platform requires the collaborative design of various participants from different industries in the city [235], to track the onsite behaviors of construction workers via computer vision to generate AI models [226]. However, it is undeniable that in future research on sustainable building, more attention should be paid to the health and safety of stakeholders, so that AI and BIM technology can assisted in achieving the goals of SDGs 3 (people’s well-being) and SDGs 17 (stakeholders).

4.3. Challenges and Limitations

The findings of this paper show the relationship between AI and BIM technologies for sustainable building in the context of smart cities, which have challenges and limitations in sustainable design, sustainable construction, sustainable development, and building life-cycle stages.

In terms of AI and BIM during sustainable design, the results of Section 3.2.1 suggest that the building material and cost are a challenge for implementing AI and BIM during sustainable design to achieve efficiency of energy and carbon emissions. In addition, there are still gaps in the reduction in water demand in building, onsite, and logistics management through the use of AI technology and BIM for sustainable design. Therefore, the integration of various cutting-edge AI technologies and BIM should be employed and embedded into sustainable design for a wide range of optimized applications for addressing the effectiveness of cost, building materials, and water management to achieve the sustainability of the building in smart cities.

In terms of AI and BIM for sustainable construction, the results of Section 3.2.2 indicate that although the emerging ITs have been increasingly deployed in integrated construction for the popular implementation of modular construction and prefabrication construction method in sustainable construction via AI technology and BIM, this method lacks efficiency when the number of building components is large, particularly those that have the greatest volume of building information for on-site management. Thus, the development of cutting-edge infrastructure and application of ITs by integrating with AI technology and BIM should be set to solve the problem of efficiency in managing massive information of building components for modern modular and prefabrication construction methods.

In terms of AI and BIM for sustainable development, the results of Section 3.2.3 highlight that there is a challenge to ensure inclusive and equitable quality education and promote lifelong learning opportunities for all (SDG 4), and promote sustained, inclusive and sustainable economic growth, full and productive employment and decent work for all (SDG 8) via adoption of AI and BIM for sustainable development. In addition, to obtaining accurate and optimized data through methods of AI and BIM for ensuring access to affordable, reliable, and sustainable modern energy for all (SDG 7) is a current drawback, and more work is needed for building resilient infrastructure, promote inclusive and sustainable industries, and foster innovation (SDG 9). Further, the present barrier is whether the shortest evacuation route is the safest in integrating the potential of neighboring buildings with the use of AI and BIM to build inclusive, safe, risk-resilient and sustainable cities and human settlements (SDG 11) through developing sustainable communities.

Hence, future work could be conducted to learning more about the benefits of a method for automating project scheduling using project BIM in educational settings to promote equal access to new technologies in education, be focused on the updating BIM data by point cloud to make the model more accurate and obtain optimized output, and develop methods for smart safety for sustainable buildings. Importantly, these challenges, drawbacks, and barriers are closely associated with stakeholders of sustainable buildings in smart cities, which currently lack sufficient studies focusing on human health and comfort, technical education and training of constructors, and opinions and information exchange between stakeholders in the construction process. Future research could focus on people-centered, healthy, efficient, and sustainable buildings in the context of smart cities integrating various emerging ITs facilitated by AI and BIM to address SDGs. In terms of AI and BIM in building life-cycle stages, the results of Section 3.2.4 reveal that although the use of integration of BIM and LCA methods assists in effectively improving building performance, there is still a challenge in generating optimized digital models for efficient design decisions by implementing AI and BIM. Surprisingly, there is a challenge for the adoption of AI and BIM across the facility management stage and restructuring and waste recovery stage. As such, AI has the potential to further BIM–LCA integration in smart cities.

However, this paper is limited to using data from the WoS core database, excluding data types such as articles and journals from non-core databases, which ensures higher-quality data sources but may lead to the omission of interesting studies related to AI and BIM technologies for sustainable building. This approach limits the scope of this research, which might not comprehensively investigate emerging technologies such as deep learning, machine learning, artificial neural networks, computer vision algorithms, evolutionary algorithms, genetic algorithms, and NSGA-II, in smart cities for sustainable building. Additionally, this paper does not validate the potential of AI and BIM technologies through empirical research to address sustainable building development challenges in smart cities. These limitations suggest opportunities for future studies to explore different databases (e.g., Scopus) to complement and segment research on AI and BIM technologies.

5. Conclusions

In this paper, the bibliometric method is adopted to study the research status and development trend of BIM technology and AI technology in sustainable building since 1998 to the sustainable development of society under the background of a smart city, and the following contributions are made: (1) By using macro and micro research methods, this paper discusses the influence status of AI and BIM in sustainable building from three aspects of sustainable design, sustainable construction and sustainable development. It is revealed that the combination of AI and BIM is conducive to the optimization decision of material, cost, energy, construction scheduling and monitoring, and promotes the development of sustainable building in both technical and human aspects. It is believed that the combination of AI, BIM and LCA technology has great prospects for improving building performance. AI and BIM-based integration of digital ITs such as IoTs, 5G, sensors and VR will further advance the development of sustainable buildings on the basis of SDGs 7, 9, 11 and 12; (2) This paper uses the bibliometrics tool Citespace for keyword co-occurrence, cluster, and burst analysis. Through keyword co-occurrence, clustering, and citation bursts, the current research hot spots and future research trends of AI and BIM in sustainable buildings can be understood, and further analysis can be made by searching literature sources in the WOS database. Compared with the current literature of the same type, there is no literature to analyze the application of AI and BIM in sustainable buildings by keyword co-occurrence, cluster, and burst analysis. In terms of keyword co-occurrence, AI, BIM, design, construction, management, system, framework, and optimization are the keywords that appear more frequently, of which AI has more connections with other keywords. In addition, currently, sustainable development, machine learning, design, system, management, model, sustainable development goal, performance, and big data are also relatively concentrated research hotspots. Whilst, machine learning, design, model, system, and big data are approaches and technologies to aid AI and BIM in sustainable building towards sustainable development, of which, since year 2018, cutting edge technologies such as big data and machine learning, have rapidly become topical technologies in the field, and have been integrated to address SDGs from 2019. Further, in the process of building development, the importance of management and building performance issues have been more concerned for sustainable building. In terms of keywords cluster, among the nine clusters, ITs with BIM and AI for addressing sustainable building are integrated with clusters such as emerging technologies, and data-driven technologies, which are focused on popular functions such as calculation tool and compressive strength clusters. Additionally, the adoption of of AI and BIM for sustainable building, which is associated with sustainable concepts including sustainable development and regarding sustainability clusters, currently is research (cluster) driven, and business model (cluster) acts a key role. In terms of citation bursts, sustainable design has a strongest citation bursts in the past. machine, future, big data, implementation, design, embodied energy, and LCA are the most recent emergent words over the last five years, of which the big data and LCA have the highest burst intensities and outbreak period of machine since year 2022 is still continuing although it has lowest strength. (3) This paper provides designers with a need for self-improvement under the digital technology based on AI and BIM and a sustainable human-centered design direction, and also provides technology developers and researchers with integrated, easy-to-use, compatible and immersive AI-BIM sustainable building metaverse development ideas.

However, there are a few challenges, drawbacks, and barriers that need to addressed: (1) building material and cost to achieve efficiency of energy and carbon emissions and reduction in water demand in building, onsite, and logistics management through the use of AI technology and BIM for sustainable design; (2) large number of building components that have the greatest volume of building information for managing on site to modular construction and prefabrication construction method in sustainable construction via AI technology and BIM; (3) generating optimized digital models for efficient design decisions and implementing AI and BIM across the facility management stage and restructuring and waste recovery stage in building life-cycle stages; and (4) insufficient research on human health and comfort, technical education and training of constructors, and opinions and information exchange between stakeholders in the construction process to achieve sustainable development goals (e.g., SDG 4, SDG 8, and SDG 9). At present, one limitation of this paper is that bibliometric research still relies on the WOS database, so it only discusses the hot spots and directions of the research from the perspective of one database. However, the method of further searching for relevant papers in the references of micro-documents further expands the scope of this research. Future research can consider using multiple different databases, such as Scopus, and extend the research to CIM and AI based on the integration of multiple technologies for sustainable smart city development centered on human health.