The Role of Artificial Intelligence in the Transformation of the BIM Environment: Current State and Future Trends

Abstract

The article explores the role of artificial intelligence (AI) in the transformation of the Building Information Modeling (BIM) environment. It focuses on current trends and developments in the field of AI, its applications in BIM, and future perspectives. AI introduces process automation, design optimization, and efficient management of construction projects into the BIM framework. Among its many benefits is the ability to quickly retrieve information, identify and describe problematic areas, and suggest alternative solutions during the design phase. In the construction phase, AI can enable more efficient and faster responses to project changes and construction challenges. When transferring data to the operation and maintenance phase, AI can sort large volumes of information and present it in a clear and actionable format for facility managers. This article analyzes both theoretical and practical aspects of AI integration into BIM and evaluates its potential benefits for the construction industry.

1. Introduction

The digital transformation of the construction industry represents a fundamental shift in the approach to designing, constructing, and managing buildings. A key component of this process is Building Information Modeling (BIM), which enables the creation and management of digital models of construction projects throughout their entire life cycle. With the increasing complexity of projects and the growing emphasis on efficiency, sustainability, and construction quality, the integration of Artificial Intelligence (AI) into the BIM environment is gaining increasing prominence. This combination is profoundly reshaping current practices and unlocking new possibilities for process automation, design optimization, and intelligent building management [1,2].

AI within the BIM framework offers significant potential in analyzing large datasets and generating optimized solutions—whether in building design, maintenance planning, or construction process management. Machine learning algorithms, for instance, enable the detection of design flaws, prediction of construction risks, and automated inspection of structural elements using computer vision [3,4]. Generative design provides architects and engineers with tools to generate numerous design variants based on parametric inputs, leading to more efficient use of both materials and space [5].

Another significant area is the integration of AI with the concept of digital twins—dynamic digital replicas of real-world structures, connected in real time through sensor networks. These systems enable continuous building condition monitoring, predictive maintenance, future scenario simulations, and operational optimization, including energy efficiency improvements [2,6]. Despite these benefits, integrating AI into BIM also faces several critical challenges—including the lack of high-quality, standardized data, software interoperability issues, and the need to expand professional competencies in data analytics and algorithm development [3,4].

This article aims to comprehensively analyze the role of AI in the BIM environment, assess its benefits, explore current trends, and identify potential directions for future technological development. Specific objectives include mapping the current state of AI adoption in BIM, including recent technological advancements; analyzing AI implementation strategies across various stages of the construction process (design, construction, operation, maintenance); identifying key barriers and proposing strategies to overcome them; and finally, forecasting future trends and the impact of AI on the digitalization and automation of the construction industry over the next decade.

To achieve these objectives, a combination of qualitative and quantitative research methods is employed. The foundation of the research is a systematic literature review and content analysis of professional articles, reports, and publications from the past five years focusing on AI and BIM. This is followed by a comparative analysis that examines approaches to AI integration into BIM across selected countries and identifies successful implementation cases. The research also includes predictive and trend analyses using available data and analytical tools to estimate future developments and identify patterns in the application of AI technologies in construction practice. The article further includes a demonstrative application of generative AI models—ChatGPT-4o and Gemini—to illustrate their potential for supporting design and decision-making processes within Building Information Modeling.

The outcome of this methodology is a comprehensive overview of the current state and future outlook of AI applications in BIM. The research aims to contribute not only to a deeper understanding of the ongoing digital transformation in construction but also to the formulation of concrete recommendations for the effective use of AI in practice.

1.1. Methodology of Review and Predictive Analysis

This review was conducted as a systematically structured literature mapping with quantitative synthesis of emerging trends. The primary sources of bibliographic records were the Scopus and Web of Science—Core Collection databases. The search covered the period 2019–06/2025, with a particular emphasis on the most recent five years in order to capture both the current state of knowledge and the latest dynamics of development. The search strategy combined controlled key terms and their synonymous variations in logical combinations: (“building information modeling” OR BIM) AND (“artificial intelligence” OR AI OR “machine learning” OR “deep learning” OR “computer vision” OR “digital twin” OR “generative design”). Queries were applied within the Title–Abstract–Keywords fields and adjusted to the syntax of the individual databases. The language was restricted to English and Czech.

The inclusion criteria covered peer-reviewed journal articles and contributions from peer-reviewed conferences within the fields of construction, architecture, and facility management, comprising both empirical studies (experimental and case-based) and review or methodological papers. Excluded were publications from non-relevant disciplines, non-English/non-Czech contributions, non-peer-reviewed preprints lacking empirical data, works without full text, and duplicates. Deduplication was carried out by matching DOI/ISBN, title, and authors.

Screening proceeded in two stages: (1) assessment of title and abstract, and (2) full-text review. Each full text was independently evaluated by two reviewers; inter-reviewer agreement was continuously monitored, and discrepancies were resolved through discussion until consensus was achieved. For each included document, structured data extraction was conducted according to a predefined coding template: context and life-cycle stage (design–construction–operation), applied AI technique (ML/DL/CV/generative approaches), linkage to BIM platform/IFC and integration procedures, metrics/KPI (e.g., detection accuracy, time/energy savings), interoperability aspects, and reported limitations.

Study quality assessment was performed using the MMAT (Mixed Methods Appraisal Tool). Studies with low quality scores were included only in the narrative synthesis and excluded from quantitative trend derivations. To reduce the risk of bias and error propagation, we applied a simple QA workflow: (a) documented data lineage and performed deduplication of sources; (b) ran consistency checks against original documents and cross-source triangulation where multiple references were available; (c) communicated uncertainty via clearly stated assumptions and scenario ranges (rather than statistical resampling); (d) required dual human review of model-assisted inferences before inclusion; and (e) maintained versioned artifacts (data snapshots, scripts, and figure exports) to ensure reproducibility and auditability.

The quantitative component focused on adoption trends of AI in BIM and digital twins. We combined time-series data of publication activity (indicator of research intensity) with industry reports on technology penetration (indicator of market adoption), from which an aggregated indicator was constructed. Short-term dynamics were smoothed using exponential smoothing, whereas long-term market saturation was modeled with logistic curves. Uncertainty was reported as 95% bootstrap confidence intervals based on 1000 replications.

To assess robustness, three sensitivity analyses were undertaken: (i) baseline penetration shift of ±5 percentage points, (ii) variation of annual growth rates within the 7–10% range, and (iii) regulatory scenarios (growth-oriented support vs. restrictive framework), modifying saturation parameters and adoption speed.

The methodology has several limitations: it relies on secondary data of heterogeneous quality and on sources employing divergent definitions of “adoption,” which may increase heterogeneity. Publication bias and incompleteness of market reports cannot be entirely excluded. These risks were mitigated through transparent reporting of search procedures, clearly defined inclusion and exclusion criteria, separation of qualitative and quantitative synthesis, and explicit reporting of confidence intervals and sensitivity scenarios.

1.2. Definition of Key Terms (AI–BIM)

This section consolidates the key terms to ensure conceptual consistency of the contribution:

• Artificial Intelligence (AI)

Artificial intelligence is an overarching domain of computer science concerned with developing methods and systems capable of performing tasks that require human cognitive abilities—learning, decision-making, pattern recognition, and prediction. In the context of construction and BIM, AI encompasses techniques that derive recommendations from data, generate design alternatives, or perform automated checking and analysis [7].

• Machine Learning (ML)

Machine learning is a subset of AI that enables computer systems to “learn” from historical data and improve performance without explicit rule-based programming. Typical applications in BIM include cost prediction, scheduling, and clash detection based on previously recorded projects [8].

• Deep Learning (DL)

Deep learning is a specialized form of machine learning based on multi-layer neural networks capable of processing complex and unstructured data (images, video, 3D scans). Within BIM, DL is applied to the automatic recognition of building elements from point clouds, material classification, or defect detection in structures [7].

• Digital Twin

A digital twin is a dynamic virtual model of a physical object or system, connected to sensors and real-time data. In construction, it functions as a “living” replica of a building: BIM provides the structural and topological framework, while AI enables predictive maintenance, operational optimization, and simulation of future behavior [9].

2. Artificial Intelligence in BIM

2.1. Building Information Modeling (BIM) and Artificial Intelligence: Context and Integration

Building Information Modeling (BIM) represents a modern digital approach to the design, construction, and management of buildings, enabling the creation and use of intelligent 3D models that contain technical, operational, and economic information. Compared to traditional 2D drawings, BIM offers greater accuracy, improved interdisciplinary coordination, and lifecycle simulation of buildings. The resulting building information model serves as a single source of truth for all project stakeholders [10].

In response to increasing project complexity, the emphasis on sustainability, and the need for more efficient management, artificial intelligence (AI) technologies are being increasingly integrated into BIM environments. AI encompasses a set of methods and tools that enable automated processing of large datasets, predictive modeling, design optimization, and decision support. This includes machine learning algorithms, computer vision, and generative design.

The synergy between AI and BIM is expanding possibilities for project management and planning; its impact, however, differs across organizations and markets depending on data quality, governance, and skills [11].

2.2. History and Development of AI in BIM

The integration of artificial intelligence into Building Information Modeling (BIM) stems from the broader development of digital technologies and automation tools in construction. While early concepts of using AI in design, simulation, and technical management can be traced back to the 1990s, practical integration of AI and BIM began to materialize only after 2010 [12,13]. This milestone is associated with the emergence of commercially available BIM platforms, advancements in computational power, machine learning algorithms, and structured data formats.

The following overview focuses on three main phases of development—from initial applications to today’s advanced systems enabling autonomous decision-making and integration with digital twins.

2.2.1. Initial Use of AI in BIM (2010–2015)

In this first phase, AI technologies in BIM focused primarily on supporting visualization and data analysis. AI was used to create predictive models that helped identify potential design issues before construction began. Significant advances were made in automated clash detection between building elements, which substantially improved interdisciplinary coordination [14].

2.2.2. Expansion of AI in Design and Project Management (2016–2021)

In the subsequent period, AI became an integral part of design and project management processes. Generative design emerged, enabling the creation of multiple design alternatives based on predefined input parameters, such as budget constraints, energy efficiency, or spatial organization.

At the same time, BIM models began to be linked with real-time data, laying the groundwork for intelligent construction scheduling and progress tracking. AI-driven algorithms were also applied to risk analysis and preventive maintenance planning, thereby increasing sustainability and operational efficiency [11].

2.2.3. Current Phase (2022–2024): Digital Twins and Autonomous Systems

In the most recent phase, development has focused on digital twins—virtual replicas of physical objects connected to sensor and IoT data. This technology enables real-time building monitoring, behavior prediction, and optimization of operational strategies.

There has also been a significant shift in the development of autonomous decision-making systems that, through a combination of machine learning, neural networks, and computer vision, allow for the automatic management of construction processes. These systems can detect structural defects, schedule maintenance interventions, and manage energy consumption according to current operational conditions [10].

AI now also supports the implementation of augmented reality (AR) and virtual reality (VR) in the BIM environment. These technologies facilitate realistic project visualization, enhance team communication, and enable interactive design review before construction begins.

2.3. Key Artificial Intelligence Technologies in BIM

Artificial intelligence (AI) encompasses a wide range of technologies increasingly applied within the Building Information Modeling (BIM) environment. The main technologies include:

• Machine Learning (ML)

Machine learning is a core component of AI that allows systems to learn from historical data without explicit programming. In the BIM context, ML is used to analyze extensive datasets from previous projects to identify recurring patterns, trends, and anomalies. A typical example is the automatic recognition of clashes between building components (e.g., pipes and beams) in 3D models, where ML algorithms not only detect errors but also suggest solutions based on previous projects.

Another application is cost or schedule prediction based on project input parameters. ML algorithms can anticipate potential delays or budget overruns before they occur, enabling timely project team interventions. During the planning phase, ML can also optimize construction sequencing or material selection based on sustainability and cost-effectiveness [14].

• Neural Networks and Deep Learning (DL)

Deep learning, a more advanced form of ML, works with multilayered neural networks capable of processing complex and unstructured data. This technology is particularly useful in BIM when working with visual or spatial data such as photographs, point clouds, or scanned data.

One example is the automatic reconstruction of BIM models from 3D scan data. Neural networks can recognize and classify various building elements (e.g., walls, windows, pipes) and convert unstructured scans into structured BIM models. This significantly accelerates the digitalization of existing buildings and supports the creation of digital twins [10].

DL is also used for classifying types of construction components, estimating material conditions, recognizing construction details, and modeling complex interactions within building systems.

• Computer Vision (CV)

Computer vision allows computers to “see” and interpret image data similarly to the human eye. In the BIM domain, CV is used for automated monitoring and control of construction activities using camera systems, drones, or mobile devices.

Typical uses include tracking construction progress by comparing visual records with the planned state of the BIM model. The system can automatically detect deviations, delays, or missing elements and alert the project team. CV can also identify potentially hazardous situations on-site, analyze the movement of people and machinery, and contribute to occupational safety [11].

Computer vision is also a key tool for quality control—e.g., detecting cracks, deformations, or poorly executed joints not easily visible to the naked eye.

AI technologies—from ML and deep neural networks to CV—significantly expand BIM’s capabilities. They enable faster and more accurate design, efficient data management, predictive risk control, and optimized construction processes. These tools not only enhance project quality but also contribute to sustainability, safety, and economic efficiency. As they continue to develop, deeper integration of AI across all phases of the building lifecycle is expected.

One illustrative example is the application of AI to the IFC standard. Analyzing this file format and identifying errors can be a key benefit. For demonstration, after uploading the IFC file “01_DiMS_Strojovna VZT_ASR.ifc” and entering the prompt “identify errors in the model” into an AI chatbot (e.g., https://chatgpt.com (accessed on 20 July 2025)), the following responses were generated:

• Duplicate names None found—no elements with the same name in multiple locations (which is positive).
• Elements without spatial placement (no coordinates)—Many elements were found (e.g., structural components, surface layers, type elements) that lack defined positions in the model. Some examples:
  • Floor: Ceramic tile 300 × 300 mm, thickness 10 mm: 1,354,902
  • Single-leaf door: 800 × 1970 mm: 1,343,408
  • Portable fire extinguisher: 1,358,041
  • Office chair: 500 × 500 mm
  • Basic wall: Ceramic blocks type THERM thickness 300 mm

These elements likely exist as type definitions or are indirectly linked, but do not contain direct placement (IfcLocalPlacement → IfcCartesianPoint).

The AI chatbot can even instantly suggest a Python (Version 3.12.5) script for visualizing the IFC model, as you can see in Figure 1.

The overview of key technologies in this chapter reflects the authors’ perception of AI primarily as a practical extender of the data capabilities of BIM, rather than as a substitute for professional judgment. Generative procedures are considered to serve for rapid exploration of design alternatives, yet only with mandatory expert verification and compliance checks against LOIN/ČSN. CV/DL are seen as having the greatest added value in comparing as-built versus as-designed models and in continuous quality control, where repeatable metrics (precision/recall) and integration with BCF are essential.

It is critically important that each AI module both ingests and outputs through interoperable interfaces (IFC 4.3, BCF), with explicit documentation of data and model provenance (data lineage), and that its recommendations remain explainable (XAI); otherwise, AI risks merely amplifying data noise and increasing operational risks. When these conditions are met, AI becomes a reliable accelerator of quality, safety, and efficiency across the entire building life cycle—from design through construction to operation.

3. Main Applications of Artificial Intelligence in BIM and Its Benefits for the Construction Industry

Artificial Intelligence (AI) brings a wide range of innovations to Building Information Modeling (BIM), significantly contributing to the automation, optimization, and digitalization of construction processes. The integration of AI with BIM enables the processing of large volumes of data, automation of routine tasks, and improvement of decision-making processes throughout the entire building lifecycle—from design to operation and maintenance. The main applications of AI in BIM and their benefits for the construction sector are outlined below.

3.1. Generative Design and Design Optimization

Generative design represents an innovative approach in architecture and construction that leverages machine learning (ML) and deep learning (DL) algorithms to create and optimize architectural designs. This process involves an AI-based design system generating a large number of layout variants based on predefined parameters such as spatial requirements, material costs, environmental impact, or energy efficiency. These variants are then evaluated and optimized, and the algorithm selects the most advantageous solutions according to specified criteria [15].

According to Altaf et al. (2024), AI-driven generative design can reduce construction material costs by up to 30% due to more efficient spatial planning. This approach allows architects and engineers to develop building project variants more quickly and to better utilize available space, minimizing material usage and reducing the environmental footprint of construction [1].

Another significant benefit of generative design is its ability to simulate various environmental factors, such as natural light availability, ventilation, and energy performance of buildings. Machine learning algorithms enable predictive analysis of energy consumption, resulting in designs that not only meet aesthetic and functional requirements but are also energy efficient [16].

AI is also capable of generating thousands of layout variants within minutes—a process that would otherwise take architects weeks or even months of manual work. This technology significantly accelerates the design phase and reduces the workload of engineers and architects, allowing them to focus on the detailed aspects of the project instead of repetitive tasks [15].

The design process of building elements can begin with a simple parametric input and later include advanced modifications of technical rooms, placement of technical equipment, furniture, and specialized infrastructure.

An example of such application is the following prompt submitted to a chatbot interface: “Create a 3D view of the layout of the first floor of a building with dimensions 30 × 20 m. Design a layout for a family house.” Based on this simple input, the system generated two alternative layout variants, see Figure 2 and Figure 3. AI generated alternative floor plan layouts. The evaluation focused on whether the floor plan complied with basic spatial logic and whether the outcome could be used as a design concept. The output was assessed primarily on a visual basis—as an illustrative demonstration of AI’s capability to generate design variants.

However, it is important to acknowledge that floor plan generation using AI has significant limitations. Outputs based on textual input often lack technical accuracy, construction logic, and compliance with regulations and standards. Currently, AI tools cannot fully replace the expertise of an architect—the generated outputs should rather be seen as inspirational concepts or quick design alternatives, not as final solutions ready for implementation. The case study further illustrates that while the results may appear visually convincing, they often require thorough review and professional corrections in terms of feasibility.

In summary, generative design proves most meaningful as a rapid exploration of the solution space, supported by clearly defined evaluation criteria (energy performance, material intensity, spatial logic). Within minutes, it can expand the number of tested alternatives and free designers’ capacity for conceptual decision-making. However, the practical applicability of these outputs depends on the quality and completeness of the input requirements and on subsequent expert verification; without this step, text-driven generation remains more of an inspirational sketch than a realizable solution compliant with regulations and construction logic.

The optimal workflow can be considered one in which generative systems are firmly integrated with BIM (IFC/BCF), deliver measurable criteria, and connect to rule-based checking and standards. Only such integration justifies the declared savings in time or material and enables a transparent selection of the final variant.

3.2. Automated Error Detection and Risk Management

In the field of civil engineering and architecture, one of the major challenges is the error rate in design and construction components, which can lead to significant financial losses and construction delays. BIM models contain vast amounts of data that can be analyzed using artificial intelligence (AI) to identify design flaws, deficiencies in structural elements, or detect clashes between building systems (e.g., electrical systems vs. piping).

According to a study by Olugboyega et al. (2024), machine learning algorithms can detect construction errors with up to 92% accuracy, enabling their early resolution and significantly reducing additional costs related to rework and corrections. Automated error detection thus represents a revolutionary step toward enhancing the quality of construction projects and reducing the likelihood of human error in the design phase [4].

Another area where AI contributes to the optimization of BIM processes is risk management during construction. Based on historical data, AI can predict potential risks and propose preventive measures, thereby minimizing potential issues. This approach has proven particularly effective in safety management on construction sites, where AI-driven systems can monitor the work environment and detect hazardous situations before accidents occur [15].

An example of such implementation can be seen in projects that combine computer vision and machine learning for real-time construction site monitoring. AI analyzes footage from cameras and drones and can automatically detect inconsistencies such as incorrect procedures, improper placement of building components, or the presence of unauthorized personnel on site [17].

In addition to design error detection and construction site safety monitoring, AI also optimizes construction planning. Based on project development simulations, it can predict potential delays and suggest alternative scenarios that help minimize the risk of schedule slippage. This approach has already proven effective in large-scale infrastructure projects, where frequent schedule changes require flexible responses to unforeseen situations [16].

Based on the design shown in Figure 2 of the previous chapter, Google’s AI chatbot Gemini was used to demonstrate the capabilities of automatic error detection. The input instruction stated: “Review the design, identify and describe errors in the model.” This approach illustrates the potential of generative AI for rapid analysis of design variants, with the goal of verifying the system’s ability to detect potential flaws in the layout without manual intervention by the designer.

The analysis revealed several key shortcomings in the design. The first was the absence of an entrance hall or vestibule, with direct access from the exterior into the living space, which negatively impacts hygiene, comfort, and privacy. The kitchen area was deemed undersized, with a capacity more suitable for a small unit than a family house of this scale.

The placement of the dining table was evaluated as ergonomically inappropriate due to its considerable distance from the kitchen, reducing user comfort. The design was also criticized for an insufficient number of sanitary facilities—despite the house having four bedrooms, there was only one centrally located bathroom, without an additional separate toilet or en-suite facilities.

Further issues were identified in the bedrooms, including inadequate size and the absence of built-in storage. The master bedroom lacked even a nightstand. A technical room or laundry area was entirely missing, even though a house of this size would typically require a dedicated space for a washing machine, dryer, and other utilities.

The AI tool also flagged a lack of general storage capacity (e.g., pantry, utility rooms, cleaning closet), issues with natural lighting and ventilation in certain interior spaces, and compromised privacy of the centrally located bedroom due to its direct access from the living zone.

Finally, the design was criticized for inefficient spatial circulation and flow, particularly due to long and narrow hallways occupying a disproportionately large portion of the floor area. These findings demonstrate that even when using advanced generative tools, the outputs must undergo expert validation. Without architectural judgment and knowledge of functional relationships, the resulting layout can be highly problematic in terms of usability.

The practical example shows that generative AI can rapidly generate spatial layouts and detect basic flaws. However, it also exposes major risks. The outputs often lack operational logic, technical accuracy, and compliance with regulations. Without expert assessment, such designs may be functionally and structurally inadequate.

Another significant issue is the lack of transparency in AI decision-making—although the AI proposes solutions, it often does so without clear reasoning, making validation and further development difficult.

Therefore, AI can serve as a useful design tool, but not as a replacement for qualified architectural judgment. Its outputs must always be critically reviewed and adjusted according to real-world requirements and professional standards. The greatest impact of AI emerges when deviation detection is firmly tied to quality management and scheduling, rather than operated as an isolated audit of the model. Its practical benefit increases particularly where the input BIM data are complete and consistent (e.g., unified element identification and up-to-date interdependencies between disciplines), since only in such an environment can AI findings be directly translated into actionable tasks with feedback to both the model and project management.

At the same time, continuous monitoring of construction progress (e.g., image analysis from cameras and drones, site risk assessment) and forecasting of schedule impacts have proven to be key, as they shorten the latency between the emergence of a problem and its resolution. To prevent automation from introducing noise, curatorial work with detection thresholds, prioritization of outputs, and expert validation before committing changes to the model and schedule is essential. In this configuration, AI becomes a stabilizing element of the project: it strengthens prevention, reduces rework and secondary costs, and increases the predictability of progress without aspiring to replace professional responsibility.

3.3. Digital Twins and Preventive and Predictive Maintenance

Digital twins represent a rapidly maturing technology in building management, enabling the creation of virtual replicas of physical construction assets that are connected to real-time sensor data. This technology allows not only real-time monitoring of a building’s condition but also the prediction of future developments and automated optimization of operational parameters. The integration of artificial intelligence (AI) into digital twins opens entirely new possibilities in the construction sector, particularly in the fields of predictive maintenance and efficient energy management.

3.3.1. Integration of Digital Twins with AI and Its Benefits

AI enables the interpretation of sensor data collected from buildings and its use in making automated operational and maintenance decisions. For example, HVAC systems (heating, ventilation, and air conditioning) can be automatically adjusted based on current conditions and predicted energy consumption, leading to significant energy savings. Berigüete et al. (2024) report that optimizing building operational parameters with AI can result in energy savings of up to 20%, which has a significant impact on sustainability and operating cost reduction [2].

3.3.2. Preventive and Predictive Maintenance and Fault Detection

One of the most important benefits of integrating AI with digital twins is the ability to predict failures and optimize maintenance. AI models can analyze historical building operation data and identify patterns that signal impending failures of critical systems. For instance, AI-driven algorithms can detect wear and tear in HVAC systems, electrical wiring, or structural components before a malfunction occurs, enabling timely preventive maintenance [18].

In their study, Lemian & Bode (2025) demonstrated that linking AI with digital twins can predict system failures with up to 87% accuracy. This means that facility managers can reliably anticipate system malfunctions and perform maintenance precisely when needed, thereby reducing operating costs and minimizing downtime [6].

AI is also highly effective at parsing manuals and large volumes of PDF documentation, which are standard components of project documentation. Manufacturers typically provide detailed instructions for preventive maintenance in extensive manuals, and extracting relevant data from these is crucial. In this regard, AI undeniably saves significant time by retrieving necessary information from operations and maintenance documentation.

To demonstrate this application,

Table 1. Example of Automatic Export from the PDF Manual 39.02.02-Manual EPS Profile 815D.pdf (authors).

Frequency	Task
Daily	Verify that the control panel indicates normal status (no LED faults/alarms, LCD displays date and time).
Daily	Confirm that all previously recorded faults have been addressed.
Weekly	Clean the front panel of the fire alarm control unit using an appropriate cleaning agent.
Weekly	Activate a device (manual or automatic detector) and verify system functionality.
Weekly	Log the use of the test device in the inspection logbook and reset the control panel.
Weekly	Check the status of printouts from printers connected to the system and replace ribbon if needed.
Weekly	Ensure an adequate supply of paper in the printers.
Weekly	Record all faults in the logbook and carry out corrective actions.

was generated automatically by an AI chatbot after importing the PDF manual 39.02.02-Manual EPS Profile 815D.pdf (available at: [https://app21.connect.trimble.com/tc/api/2.0/s/RW-VekLXZe_tZftQS0-NoZNC5RQpOsFNKqxtFVFG_fgJL85-Z2Ejh1kgKUJSAjI2] (accessed on 20 July 2025)) for an 18-page fire alarm system manual. The task was to create a table of regular inspections and maintenance activities. The document defines these inspections and maintenance tasks in various sections, and AI extracted and organized them accordingly. As you can see in Table 1.

3.3.3. AI Combined with IoT for Building Management

In addition to analyzing historical data, digital twins are increasingly integrated with IoT (Internet of Things) sensors, which enable continuous data collection on the condition of the building. For example, sensors can monitor humidity, temperature, structural loads, energy consumption, and other parameters, which are then evaluated by AI and used for automated adjustments to building operations [19].

The use of AI-driven digital twins has proven particularly effective in industrial and commercial buildings, where operational and maintenance costs are high. Predictive models allow for the optimization of the life cycle of building components, thereby extending the building’s lifespan and reducing costs related to replacements and repairs.

The integration of IoT with BIM models is now a standard process in buildings where real-time monitoring is implemented. In this sector, AI can evaluate data and predict potential issues within the building environment.

The model shown in Figure 4 demonstrates this application on a coworking and community space. Sensors are placed in the basement of the building to test the environment within a “digital shadow” concept. The data is monitored in real time and can be configured, for example, to send notifications when there is a significant drop or rise in temperature. In this case, AI can independently assess situations and either inform the users of the building or send commands directly to the building’s measurement and control system to adjust the temperature. The model of the coworking building was supplemented with IoT sensors installed in the basement. The task was to configure real-time monitoring of temperature data within the so-called “digital shadow” concept. The evaluation focused on whether the system was capable of correctly visualizing the data within the BIM model and detecting anomalies (e.g., sudden drops or increases in temperature). It was further verified that AI could, on the basis of these data, automatically notify users or issue a command to the measurement and control system.

3.4. Automation of Scan-to-BIM Processes

The Scan-to-BIM process represents a significant advancement in the digitalization of the construction sector, enabling the automatic conversion of physical construction objects into digital BIM models. This approach relies on laser scanning of existing buildings, resulting in what is known as a point cloud model, which contains detailed information about the geometry and structure of the building.

The integration of artificial intelligence (AI) into the Scan-to-BIM process brings substantial improvements, particularly in the area of automated classification of building components. Traditional manual methods for converting point clouds into BIM models are time-consuming and require extensive human input, increasing the costs of digitalizing existing buildings. AI enables automatic identification of building elements, significantly speeding up the process and improving its accuracy [20].

According to Alshibani (2024), deep learning (DL) algorithms achieve up to 95% accuracy in converting point clouds into BIM models, resulting in considerable time and cost savings compared to traditional manual modeling methods [12]. These algorithms can automatically recognize materials, structural elements, and geometric features of objects, which facilitates building renovation and the modernization of historical structures [21].

In addition, AI enables intelligent classification of construction materials, making it easier to automatically categorize different building components. For example, neural networks can determine the material composition of objects based on visual and spectral analyses, supporting the preliminary assessment of building structures and the planning of renovation works. This approach is particularly advantageous in the restoration of heritage buildings, where detailed analysis of original materials and construction elements is crucial [20].

In summary, the entirety of Section 3.3 demonstrates the digital twin as an operational interface which, in combination with AI, transforms building management from reactive interventions to predictive maintenance—from timely detection of degradation to energy optimization and planned shutdowns. Maximum added value arises when the “loop is closed”: sensors continuously feed the model, algorithms assess conditions, and recommendations are systematically transferred into BMS/operations management, including feedback logging of interventions and their audit trail linked to the affected elements in BIM/AIM.

Practical performance is further enhanced by the automation of documentation handling—machine parsing of operation manuals into structured maintenance plans enables predictive forecasts to be linked with specific tasks, deadlines, and responsibilities. At the same time, the reliability and scalability of the solution rest on infrastructure: high-quality and stable telemetry, standardized mapping between operational data and IFC, and sufficient computational capacity; without these, latency and operating costs increase, while return on investment declines.

3.5. Intelligent Management of Construction Projects

Effective management of construction projects is a key success factor in the building industry, where constant optimization of construction schedules, supply chain coordination, and material logistics is essential. AI-driven construction management represents a new approach to project planning and execution, enabling faster decision-making based on the analysis of both historical and real-time data.

According to Ogundipe (2024), artificial intelligence enables optimization of construction schedules, resulting in a 15–20% reduction in overall construction time [22]. AI uses predictive analytics to identify potential delays and proposes alternative scenarios for more efficient use of labor and material resources. This helps minimize delay-related risks and enhances the efficiency of supply chains [21].

AI-driven tools also enable site monitoring through autonomous navigation robots that automatically compare construction progress with planned BIM models. These autonomous systems utilize computer vision (CV) and machine learning to identify discrepancies in real time and flag potential errors before they are completed. This approach significantly reduces the number of construction defects and the costs associated with their correction [20].

Another important area of AI application in construction project management is the optimization of supply chains. AI facilitates prediction of material consumption and optimization of logistics, minimizing waste and reducing the carbon footprint of construction activities. By integrating with IoT sensors, AI can track the current status of material inventories, enabling just-in-time deliveries and minimizing excessive storage of building components [22].

Building on the examples presented, project management can be understood as a domain where AI delivers the greatest impact when tightly coupled with BIM/CDE and operating on validated WBS/BoQ data as well as real-time sensor and logistics inputs. In such a configuration, algorithms do not serve as an “autopilot,” but rather as a coordinated orchestrator that continuously recalculates schedules, capacities, and supplies, presenting feasible scenarios with a transparent decision trail. Properly established workflows of this kind can in practice shorten construction timelines and reduce defects, with the primary benefit stemming from the early identification of critical paths, resource conflicts, and supply chain optimization (just-in-time delivery, minimization of downtime).

At the same time, the performance of these tools is limited by the quality of inputs and the degree of integration—without standardized data, clearly defined constraints (working windows, safety, technologies), and connections to procurement and logistics, predictions quickly deteriorate and efficiency gains dissipate. The optimal configuration is therefore a “closed loop”: AI generates a draft plan, which is validated by an expert and approved in the CDE; execution produces new evidence (progress, consumption, delays); and the model is re-optimized in short cycles. Only through this process can the declared time and cost savings be translated into repeatable, auditable practice.

4. Recent Technological Advancements in AI-Supported BIM

As previously discussed, recent years have seen significant progress in the technological capabilities of artificial intelligence (AI) in the field of Building Information Modeling (BIM). The growth of computing power, the availability of cloud platforms, and advancements in augmented reality (AR) and virtual reality (VR) have all contributed to increased efficiency in construction projects and improved lifecycle management of buildings. The following chapter focuses on the latest trends in AI and BIM, including their applications in cloud-based BIM platforms, AR/VR, and predictive modeling.

4.1. Integration of AI into Cloud-Based BIM Platforms

Modern BIM software, such as Autodesk BIM 360, ArchiCAD 28, and Bentley OpenBuildings Designer CONNECT Edition 2024, Update 10.1 is increasingly being integrated with AI-driven tools that enable process automation, clash detection, generative design, and predictive maintenance. Cloud-based BIM platforms enhanced with AI allow real-time data sharing and analysis, resulting in more efficient collaboration among project teams.

According to Nasiri (2024), Autodesk introduced an AI-driven module for automatic error detection in models in 2023, significantly reducing the need for manual reviews [23]. This module uses machine learning to identify inconsistencies in construction projects and provides suggestions for optimizing building design [23].

Another key benefit of cloud-based BIM platforms with AI is their ability to connect with IoT sensors. This integration enables automated building management and monitoring, where AI analyzes sensor data and proposes energy-efficient solutions [24]. As a result, construction companies can reduce operational and maintenance costs by up to 25%.

From the above, it follows that cloud-based BIM platforms enhanced with AI are meaningful only when they function not merely as a “repository with analytics,” but as the backbone for change management: from automatic error detection in models, through the generation of design suggestions, to the integration of operational data and maintenance interventions. The practical benefits increase with the degree of integration into the IoT ecosystem, where AI outputs are not stored in isolated dashboards but instead trigger concrete decision-making and operational actions on the shared model (AIM)—with measurable impacts on costs and energy performance.

For such a scenario to be scalable, it is essential to rely on open interfaces and standards (IFC/BCF, data requirement specifications), rigorous version control, and auditability of model recommendations; without these elements, the advantages of the cloud are quickly offset by operational costs, latency, and vendor lock-in. At the same time, it is advisable to establish clear rules for input data quality and their curation, since these factors determine whether the promised automation translates in practice into shorter review cycles, reduced clashes, and more stable operation across the entire building life cycle.

4.2. AI for Augmented and Virtual Reality (AR/VR)

The combination of AI, BIM, and augmented or virtual reality enables the creation of interactive visualizations of construction projects and simulations of work processes. AI-driven VR models allow for the simulation of various building behavior scenarios before construction begins, helping architects and engineers optimize designs at early stages [25].

For example, AI-powered AR assistants on construction sites allow workers to precisely place building components according to the digital model. AR devices provide on-site visual instructions, reducing installation errors and increasing productivity.

With the help of AI, VR models can also be connected to sensors, enabling real-time monitoring of structural changes. This allows project managers to track construction progress, simulate potential risks, and optimize processes before they are physically implemented [26].

The presented examples indicate that AR/VR enhanced with AI provide the greatest value when functioning as interfaces for decision-making and execution—rather than merely as presentation “showcases.” Their benefits materialize only when overlays and simulations are firmly linked to the versioned BIM/AIM model, and outputs (findings, instructions, changes) are fed back into the CDE in the form of BCF tasks and execution records.

Practical applicability, however, depends on precise geometric registration (tolerance to alignment errors), low latency, and proper device calibration; without these, element displacement and contextual misinterpretation on site become risks. In execution, such configurations result in reduced rework and faster assembly, while in design they enable clearer interdisciplinary coordination and earlier clash detection. To avoid the “attraction effect,” we recommend that AR/VR workflows be assessed against measurable KPIs (task time, error rate, rework, stakeholder comprehension); otherwise, the declared benefits remain difficult to transfer into practice.

4.3. AI-Driven Predictive Modeling and Simulation

AI-driven predictive modeling enables the creation of advanced simulations of construction processes, helping to minimize risks, optimize material usage, and improve overall project efficiency.

According to studies [27,28], AI models can predict the behavior of construction materials and simulate their degradation under various climatic conditions. This approach supports the design of more sustainable structures and helps prevent wear and failure of critical building components.

Another application of AI in predictive modeling is supply chain optimization. Machine learning algorithms can analyze pricing trends of materials and forecast optimal purchasing times, reducing costs and minimizing waste [28].

In connection with a previously generated floor plan, a demonstration was created using the ChatGPT-4o tool to simulate interior daylighting. The tool combines AI and graphical calculations to visualize the impact of natural light on indoor spaces. The assessment focused on how the tool visually represented the impact of daylight on the interior and whether the image allowed for evaluation of window placement, shading depth, and lighting quality. The result was considered an indicative illustration rather than an exact calculation, as you can see in Figure 5.

In this case, the simulation was set to a west-facing sun orientation at sunset, allowing for a visualization of daylight quality in selected rooms during the evening hours. Although the output is only an illustrative approximation, the tool allows the input of precise parameters such as time, date, geographic location of the building, or orientation of the structure, thereby adapting the results to specific requirements.

This type of visualization can be especially useful in the early design phase as a quick check of window placement, shading depth, or functional zoning of spaces in relation to lighting conditions. Its main advantages are speed and ease of use, without the need for complex modeling or calculations.

However, it is important to emphasize the fundamental limitations of this approach. The resulting simulation is not physically accurate—it does not account for actual climatic conditions, atmospheric exposure, the diffuse component of daylight, or the material properties of surfaces. It should therefore not be considered a valid substitute for specialized tools such as DIALux, Velux Daylight Visualizer, Ladybug, or Autodesk Insight, which provide precise values for light intensity, sunlit areas, or exposure duration.

Another limitation is the visual abstraction of the output—AI does not provide measurable data but only a graphical representation, which may be interpreted subjectively.

For these reasons, this type of AI simulation should be understood as a supplementary tool for architectural decision-making, not as a basis for assessing daylight hygiene standards or designing daylight optimization according to applicable regulations.

5. Methods of AI Implementation in BIM

Artificial intelligence (AI) is fundamentally transforming the processes of design, construction, operation, and maintenance in building projects through Building Information Modeling (BIM). Implementing AI in BIM leads to workflow automation, increases modeling accuracy, and optimizes construction process management [29]. This chapter focuses on specific methods of AI implementation across different phases of the construction process and evaluates its impact on BIM efficiency, accuracy, and sustainability.

Application of AI in Various Phases of the Construction Process

AI is applied throughout the entire lifecycle of a construction project—from design through execution to building operation and maintenance. Each phase utilizes specific AI capabilities to improve efficiency, reduce costs, increase safety, and support data-driven decision-making. In combination with BIM technology, AI becomes a key tool in the digital transformation of the construction industry.

In the design phase, AI is mainly used through generative design, where algorithms create optimized layout and structural variants based on predefined input parameters. This significantly accelerates the design process while improving output quality, thanks to AI’s ability to detect errors or inefficient solutions in the early stages. According to Yavan (2025), the use of AI-driven generative design can reduce design time by up to 50%, while material optimization can cut construction material costs by approximately 20% [30]. This allows designers to focus more on creative and conceptual decisions, while repetitive calculations and design iterations are handled by intelligent systems.

During construction, AI is applied in monitoring progress, predicting delays, and managing construction processes. The combination of machine learning with real-time data from sensors and drones enables progress tracking against the schedule, detection of risk areas, and quality control of completed work. AI also supports the automation of inspection processes, for instance, through computer vision that compares on-site conditions with the digital model. A study by Ajirotutu et al. (2024) showed that applying AI for real-time predictive analytics can significantly reduce the risk of project delays [29]. Research by Anjum & Luz (2024) further demonstrated that integrating AI and IoT sensors into safety protocols led to a 30% reduction in workplace accidents [31]. These benefits show that AI not only increases productivity but also contributes to a safer and more efficiently managed construction environment.

In the operational phase, AI enables intelligent building management through predictive analysis, simulation of operational scenarios, and optimization of energy performance. Using machine learning, energy consumption can be monitored and evaluated in real time, and building operation can be adapted to current conditions and user behavior. Yitmen et al. (2025) demonstrated that applying AI to HVAC system models resulted in a 25% reduction in energy consumption. A key component in this phase is the digital twin—a dynamic digital replica of the building connected to sensors and BIM models, enabling continuous monitoring, simulation, and real-time operational control [32]. Lemian & Bode (2025) report that this technology can reduce operational costs by up to 15% while improving transparency and flexibility in building management [6].

In the maintenance phase, AI is primarily used for predictive maintenance, where it can forecast the need for intervention before a failure occurs by analyzing historical data and sensor measurements. This approach minimizes downtime, reduces costs, and extends the lifespan of both equipment and the building itself. Huan (2024) showed that AI-based predictive approaches decrease emergency interventions and improve asset management efficiency [33]. Ucar et al. (2024) reported that the use of AI in this area led to a 20% reduction in maintenance costs. The integration of BIM with ERP systems is also gaining importance, as it enables more efficient repair planning, supplier management, and cost control within facility management [34].

Thus, the application of AI in various stages of the construction process brings significant benefits in terms of efficiency, prediction, safety, and sustainability. AI not only supports better decision-making but also transforms the very nature of the construction process into a data-driven, adaptive, and proactive approach to managing both the construction and the entire lifecycle of a building. As you can see in

Table 2. Overview of AI Applications in Different Phases of the Construction Process (Data source and derivation, see Section 5, authors).

Phase	Main AI Applications	Benefits
Design	Generative design, layout and structural optimization, design variant analysis	Faster design process (by 50%), reduced material costs (by 20%), fewer errors
Construction	Progress monitoring, delay prediction, quality control, safety analysis	Reduced accident rate (by 30%), improved project management, fewer delays
Operation	Energy optimization, digital twins, real-time system control	Reduced energy consumption (by 25%), lower operational costs (by 15%)
Maintenance	Predictive maintenance, sensor data analysis, service intervention management	Lower maintenance costs (by 20%), extended building lifespan, better planning

.

Although this chapter highlights the wide range of benefits associated with the use of AI across different phases of the construction process, it is important to note that its practical application remains selective and often depends on the specific project context, the technical readiness of the organization, and the availability of high-quality data. Many examples cited in academic literature are still pilot studies or experimental implementations, and their transferability to everyday practice may be limited. Moreover, the automation of decision-making faces the challenge of requiring expert validation—particularly in cases where AI generates outputs without explainability or transparency. Without critical oversight, there is a risk of misinterpretation or overestimation of results.

Therefore, AI should not be viewed as a replacement for human expertise but rather as a complementary tool whose benefits depend on proper integration into decision-making processes. This chapter thus provides a conceptual overview of AI’s potential, while its practical application requires thoughtful integration, interdisciplinary collaboration, and critical evaluation of outcomes—topics that are addressed in greater detail in the following sections of the text.

6. Challenges and Barriers to AI Integration in BIM

While the integration of artificial intelligence (AI) into Building Information Modeling (BIM) offers many advantages, several challenges continue to hinder its widespread adoption. These obstacles can be grouped into three main categories: economic, technical, and legislative aspects. This chapter focuses on the key barriers in the interaction between AI and BIM and proposes strategies to overcome them.

6.1. Barriers to the Mass Adoption of AI in BIM

6.1.1. Economic Barriers

One of the primary obstacles to the widespread implementation of AI in BIM is the high financial cost. Acquiring specialized software and hardware necessary for effective AI use in BIM represents a significant investment—particularly challenging for small and medium-sized enterprises. These organizations often lack the budgetary capacity to cover such costs, making AI technologies economically inaccessible [29].

Another major factor is the high cost of professional training. To use AI effectively, employees must acquire new skills in data analytics, algorithmic modeling, and working with advanced software tools. However, the costs associated with staff education and retraining pose a substantial financial burden, which may limit smaller organizations in deciding whether to adopt AI [35].

In addition to the high upfront costs, there is also the issue of insufficient short-term return on investment. Companies may not see immediate benefits from AI and are concerned that implementation costs will outweigh the potential savings and efficiency gains. This leads to hesitation and reluctance to invest in these technologies—especially in more conservative sectors less open to innovation [36].

In summary, although AI offers significant benefits for BIM, its economic accessibility remains a fundamental barrier. This could be addressed through government support, subsidies, or the development of more affordable solutions. Without such measures, the widespread adoption of AI in BIM may remain confined primarily to large companies that can afford long-term investment in these technologies [37].

6.1.2. Technical Barriers

A major hurdle to broader AI deployment in the BIM environment lies in technical limitations that complicate the integration of AI into existing design and operational systems. These issues relate to both technological infrastructure and data quality and software compatibility.

A critical problem is the lack of compatibility among different BIM platforms. A unified standard for integrating AI modules across various software systems has yet to be established. This fragmentation hinders system interoperability and often requires costly customizations or the development of proprietary interfaces [38].

Another challenge is data quality and availability. AI models rely on large volumes of accurate, current, and structured data. In practice, however, BIM models are often incomplete, error-prone, or lack standardization—significantly reducing the effectiveness of AI tools [31]. For example, automated design error detection can reach 92% accuracy, but only when the input data is complete and of high quality [39].

Computational demands are also a crucial factor. Advanced AI models operating in real time—such as those used for clash detection, maintenance prediction, or building behavior simulation—require substantial computing power. This can be limiting, especially for smaller organizations lacking sufficient hardware infrastructure or access to cloud computing services [40].

To illustrate technical complexity, consider AI-connected digital twins, which require comprehensive sensor networks and stable data streams. Ensuring these conditions is not only technically demanding but also a significant financial investment [11].

These barriers show that successful AI implementation in BIM is not just a matter of software capabilities. It requires a systemic approach involving data readiness, standardization, computing capacity, and interdisciplinary collaboration.

6.1.3. Legislative and Ethical Barriers

AI integration in BIM also faces major legal and ethical challenges in addition to technical and organizational ones. The legal framework governing AI use in construction is highly fragmented across countries. While some nations (e.g., Singapore, the Netherlands, and the UAE) actively prepare standards and guidelines for digital design tools and automation, most European countries are still in the early stages of defining regulatory frameworks. No unified legal standard currently governs the specific use of AI in architecture, construction, or building management.

For example, in the Czech Republic, the legal situation remains ambiguous. The Building Act No. 283/2021 Coll. and related legislation (e.g., Act No. 134/2016 Coll. on Public Procurement) do not account for AI usage. Legal responsibility for design, construction, and building operation rests entirely with licensed professionals, even though decisions may be significantly influenced by algorithms or generative design tools. Legally, AI is treated merely as a tool, without defined status, regulation, or liability [41,42].

A particularly problematic area is legal responsibility in the event of AI-induced error. It remains unclear who is liable—the designer, the software developer, the investor, or the system operator? In 2023, Europe saw its first court case addressing responsibility for a flawed construction design generated by AI [43,44]. This highlights the absence of a legal framework to address such incidents.

Copyright is another area lacking clear regulation. It is not defined who holds the intellectual property rights to a design created by AI—the user, the algorithm’s creator, or no one. In the Czech Republic, the law states that only a natural person can be an author—not a machine. Yet, when a design is generated from vague inputs, the questions of authorship and liability become ambiguous.

Another serious risk is the protection of personal and sensitive data, as BIM models often include detailed building information—such as security systems, operating schedules, or access mechanisms. Processing such data using AI tools can violate data protection laws (e.g., GDPR) if proper safeguards are not implemented [45,46].

These legislative shortcomings are closely linked to ethical concerns about transparency in decision-making, professional accountability, and preserving the autonomy of architects and designers. Many AI systems operate as “black boxes,” with no clear explanation of how outputs are generated. This lack of explainability is ethically problematic—especially in the design of public or security-sensitive facilities—where the designer should be able to provide a professional and rational justification for every decision, which is often not possible without a deep understanding of the AI model.

Another ethical issue is the erosion of professional responsibility, where decision-making may be delegated to an algorithm without thorough evaluation of its output. This risks shifting toward an overly passive approach (“AI made the calculation, I just used the result”). Additionally, since AI systems rely on historical data, there is a danger of perpetuating past errors, inefficiencies, or design biases—especially problematic in public projects.

In summary, without a clear legal framework and ethical guidelines, it is difficult to establish trust—both within the professional community and the broader public—regarding AI use in construction. Legal uncertainty surrounding liability, authorship, and data protection poses a significant barrier to wider implementation of these technologies. Developing methodologies, regulations, and tools for validating AI-generated outputs will be essential to making AI a trustworthy and legitimate part of the building design and delivery process. As you can see in

Table 3. Legislative and Ethical Barriers to the Use of AI in BIM (Data source and derivation, see Section 6.1, including subsections, authors).

Barrier	Type	Consequence	Note/Context
Lack of AI regulation in design	Legislative	Legal uncertainty	No unified framework in the EU; the AI Act does not yet address the specifics of the construction sector
Unclear liability in case of AI error	Legislative	Risk of legal disputes	Early cases in Europe highlight the need for clear responsibility assignment
Laws do not account for AI in construction	Legislative	Limited use in public procurement	For example, Czech Act No. 283/2021 Coll. completely omits the role of AI
Inadequate data protection in AI-processed BIM models	Legislative	GDPR violations, risk of information leaks	BIM models often contain operational or security-related data
Risk of AI misuse without understanding the outputs	Ethical	Compromised building safety, flawed decisions	“Black box” algorithms lacking output transparency
Distrust from professional bodies (ČKAIT, ČKA, RIBA, AIA)	Ethical	Resistance to innovation, slowed digital transformation	Criticism from professional chambers in the Czech Republic and abroad
Lack of a framework for validation and auditing of AI outputs	Ethical/Legislative	Low quality control, risk of abdicated responsibility	No standards for verifying quality of AI-generated designs
Unclear copyright for AI-generated content	Legislative	Ownership disputes, loss of authorial responsibility	In the Czech Republic, only a human can be an author—AI outputs are legally problematic

.

6.2. Strategies for Overcoming Challenges and Ensuring Effective Implementation of AI in BIM

The successful and sustainable adoption of artificial intelligence in Building Information Modeling (BIM) cannot be reduced to a purely technological issue. AI introduces a fundamental shift in how buildings are designed, constructed, and managed. It therefore requires a systemic and interdisciplinary approach that considers not only economic and technical factors but also legal frameworks, professional responsibility, and public trust.

The following strategies outline potential pathways to address current barriers. However, it is important to recognize that their practical application is not without risks—and critical reflection is essential for the responsible and meaningful deployment of AI in BIM environments.

One of the most frequently cited barriers is the high upfront cost of acquiring, operating, and maintaining AI systems. For small and medium-sized enterprises (SMEs) in particular, advanced digitalization may be virtually unattainable without external support. Targeted grant programs from national and EU funds could serve as an effective tool to finance software licenses, training, or access to cloud-based solutions [38].

One example of an accessible model is the concept of AI-as-a-Service (AIaaS)—the provision of artificial intelligence as a cloud-based service. This model allows access to powerful tools without requiring investment in computing infrastructure [47]. The benefits include flexibility and scalability, though drawbacks may include dependency on external providers, loss of data control, and potential security concerns.

Another significant opportunity lies in the development of open-source platforms focused on AI in BIM. These would reduce software costs, enable more transparent development, and promote collaboration between research and professional communities [48]. However, it is important to critically note that open-source projects often suffer from a lack of long-term support, irregular updates, and limited accountability for errors—factors that are especially problematic in construction applications.

From a technical standpoint, it is essential to address issues related to data compatibility and computational demand. The standardization of data formats in BIM is a prerequisite for interoperability between different AI tools. Currently, BIM practice suffers from fragmented standards, making broader integration of AI difficult. What is needed is an international agreement on data as infrastructure, including new open formats for data exchange [38,49].

Another strategic direction involves optimizing algorithms and exploring alternative computing methods, such as quantum computing, which may significantly reduce hardware demands in the future [36]. In practice, however, quantum computing is still in the developmental stage, and its real-world application in BIM remains unlikely in the near term.

A critical component is enhancing digital literacy among professionals—from designers and site managers to facility managers. It is not enough to passively use AI tools; professionals must understand their logic, limitations, and context. The introduction of training programs, certifications, and integration of AI into university curricula is essential. However, professionals already face significant pressure in terms of legislation and professional standards, which may limit their time and capacity for upskilling [50].

To ensure legal certainty, it is necessary to create a clear and predictable regulatory framework for the use of AI in design and building management. This framework should define accountability for decisions made by algorithms, establish rules for output validation, and delineate the boundary between AI’s role and the responsibility of authorized professionals [45]. Without such a framework, there is a risk of either excessive regulation (stifling innovation) or a legal vacuum that enables misuse or abdication of responsibility.

It is equally important to address cybersecurity of BIM data, particularly when AI tools operate on external servers. Protecting data, ensuring its integrity, and preventing misuse are fundamental prerequisites for building trust among professionals and the general public [51].

Finally, ethical considerations must not be overlooked. AI must not become a tool that overrides professional judgment without the possibility of explanation or critique. Transparency in algorithmic decision-making, the ability to review results, and accountability must remain intact in the era of intelligent tools. Otherwise, AI risks devaluing professional expertise and eroding trust in its outputs.

These proposed strategies show that overcoming barriers to AI implementation in BIM is a multidimensional challenge. No single measure is sufficient on its own—each must be seen as part of an interconnected system and coordinated across government agencies, professional organizations, researchers, and practitioners.

It is also necessary to acknowledge the potential side effects of some strategies—for example, centralized cloud services may threaten data sovereignty; open AI models may lead to the uncontrolled spread of unvalidated tools; and unregulated training may result in pseudo-qualifications.

Thus, the effective implementation of AI in BIM is not just a technological or economic task—it is a matter of trust, responsibility, and professional culture. This is why the process must be approached cautiously, systematically, and in dialogue with all stakeholders.

7. Forecasting Future Trends in AI and BIM

With the increasing digitalization and automation of the construction industry, artificial intelligence (AI) is expected to play an increasingly significant role within Building Information Modeling (BIM). Anticipated developments suggest that AI technologies will become an integral part of the transformative processes leading to more efficient, sustainable, and transparent management of construction projects and building operations. The following section presents forecasted development directions and key areas where AI is expected to have a major impact on the BIM environment over the next decade.

7.1. Expected Trends in AI for BIM Based on Predictive Modeling

Based on current scientific research, industry analyses, and predictive models, several key directions can be identified for the future integration of AI in BIM. These trends reflect both technological advancements and the growing demands for efficiency and sustainability in construction.

The projections are based on aggregated time series (publication activity, industry adoption surveys, market reports) and are estimated using a logistic function with 95% bootstrap intervals (1000 replications). The baseline scenario assumes an initial penetration of digital twins in commercial buildings of 20% in 2024, an annual growth rate of 7–10%, and regulatory support for energy efficiency. Sensitivity analyses vary the initial state by ±5 percentage points and the growth rate by ±3 percentage points.

One of the most prominent trends is the development of AI-powered generative design. This technology enables the automated generation of architectural and technical designs based on defined input parameters, with ongoing improvements in accuracy, speed, and flexibility. It is plausible that by 2030, around 60% of construction projects will actively utilize AI in building design, in leading markets, while adoption may be lower in SMEs and regions with limited data readiness. The development of advanced algorithms will also enable the automated creation of BIM models directly from scan data or other predictive techniques.

Another critical area of development is digital twins integrated with AI. This technology enables continuous real-time monitoring of buildings and predictive maintenance, leading to significantly reduced operational costs and enhanced building management efficiency. In optimistic in 2030, up to 80% of new commercial buildings are expected to be equipped with AI-linked digital twins, Actual rates will depend on regulation, procurement models, and interoperability progress. This approach will allow building managers not only to plan maintenance more effectively but also to optimize operational costs, particularly through intelligent energy and systems management. These changes will be evident not only in terms of operational efficiency but also in the context of sustainability, where energy consumption in buildings is expected to decrease by up to 30%.

In the construction execution phase, a rise in autonomous AI systems for construction monitoring is expected. By 2032, up to 50% of large-scale construction projects will be routinely monitored using autonomous drones and computer vision systems, overseeing the quality of ongoing work and eliminating deficiencies in real time. At the same time, it is projected that by 2035, AI-driven robotic systems will handle approximately 40% of manual tasks on construction sites, thereby increasing productivity and safety.

AI has the potential to play a major role in the context of sustainable development. AI-driven systems will enable the optimization of supply chains, reduction of the construction sector’s carbon footprint, and more accurate forecasting of material resource needs. By 2035, AI is expected to contribute to a 20% reduction in the carbon footprint of construction. Additionally, AI technologies will allow energy consumption in buildings to be optimized through predictive models and automated control of lighting, heating, and ventilation systems, potentially reducing consumption by up to 30%.

A forecast of digital twin adoption in the BIM environment was developed using exponential trend analysis based on a combination of key factors. The growth estimate took into account the current state of digital twin adoption in 2024, which stands at approximately 20% of commercial buildings, with the most common applications found in large infrastructure projects and pilot initiatives focused on building management. The projected year-over-year growth rate was derived from current trends, ranging between 7–10%, while also accounting for technological, economic, and legislative factors.

Key technological factors include improved performance of AI tools, expansion of cloud computing, and gradually decreasing costs of hardware and software. Economic and legislative factors include increasing emphasis on sustainability, efficient building operation, and mounting pressure to digitize in order to meet environmental targets and optimize operating costs.

The results of this forecast are summarized below in

Table 4. Estimated Growth of Digital Twin Adoption in BIM. (Data source and derivation, see Section 1.2 and Section 7.1; authors’ calculations) Interpretation note. The percentages reported here are scenario-based and sensitive to policy, market cycles and organizational capacity; they should not be read as point forecasts applicable across all contexts.

Year	Share of Buildings with Digital Twins (%)	Comment
2024	20	Current state of digital twin adoption in commercial construction. Mainly used in large infrastructure projects.
2026	35	Digital twins are expected to expand into more buildings, especially in the field of facility management.
2028	50	Half of new commercial buildings will include a digital twin due to reduced costs of AI and cloud computing.
2030	65	Mandatory regulations on sustainability and digitalization will drive implementation in most new buildings.
2032	75	Standardization of BIM and AI will enable effective integration of digital twins with real-time building monitoring.
2035	85	Nearly all new commercial and industrial buildings will use digital twins for operational management and optimization.

, which shows the expected proportion of buildings equipped with digital twins at selected milestones through 2035 and briefly comments on the main growth drivers for each period.

Another major area of development is automated construction inspection using AI, which is expected to become an integral part of construction project management in the coming years. AI-powered computer vision systems will increasingly be used for continuous site supervision and real-time construction progress monitoring. These technologies will not only enable faster detection of errors and discrepancies but also contribute to more efficient quality management. According to forecasts, by 2032, approximately 50% of large-scale construction projects will use autonomous AI drones for quality inspection of construction work. At the same time, it is expected that by 2035, AI-driven robotic systems will take over up to 40% of manual tasks performed directly on construction sites, marking a fundamental shift in how construction work is carried out.

Artificial intelligence will also play a crucial role in sustainable construction, primarily through the optimization of material and energy resources. AI tools will allow for more efficient supply chain management and more accurate prediction of material needs, potentially reducing the carbon footprint of construction by up to 20% by 2035. Another important application will be AI-driven models for predicting building energy consumption, which, through automated management of lighting, heating, and ventilation, could reduce energy consumption by 25–30% [52]. These approaches will not only support the achievement of environmental goals but also help reduce operational costs and improve building management efficiency throughout the lifecycle.

The methodology behind the predictive modeling is based on trend analysis and extrapolation of historical data from the past 5 to 10 years, capturing the gradual integration of artificial intelligence (AI) in BIM and the construction sector. The predictions took into account technological, legislative, and economic factors influencing the development of these technologies. Historical data show a clear upward trend in AI adoption across various areas of the construction sector, mirroring advancements in technology, the introduction of new legislative frameworks, and increasing pressure for economic efficiency in construction processes.

The prediction also reflects key technological trends, particularly improvements in computing power, which support broader real-time AI deployment, the development of IoT networks and sensor systems facilitating construction digitalization, and the growing use of AI-driven robotics, which is gradually reducing the need for manual labor on construction sites.

Based on these factors,

Table 5. Estimated Impacts of AI on BIM and Construction by 2035. (Data source and derivation, see Section 1.2 and Section 7.1; authors’ own calculations) Interpretation note. The percentages reported here are scenario-based and sensitive to policy, market cycles and organizational capacity; they should not be read as point forecasts applicable across all contexts.

Area	AI Impacts by 2035	Additional Explanation
Building Design	60% of projects will use AI for generative design	AI generative design will enable faster and more accurate building modeling. By 2035, it will be standard practice in over 60% of architectural projects.
Digital Twins	85% of commercial buildings will have AI-driven digital twins	Digital twins integrated with AI will optimize building operation and maintenance. By 2035, 85% of commercial buildings will be managed using real-time models.
Construction Monitoring	50% of construction sites will be monitored by autonomous AI drones	Autonomous AI drones will oversee construction progress and perform quality inspections. By 2035, 50% of construction projects will use this monitoring method.
Construction Robotics	AI-driven robots will perform 40% of construction tasks	AI-guided robotics will be applied in prefabrication, assembly, and other construction processes. By 2035, 40% of these tasks will be fully automated.
Energy Efficiency	30% reduction in energy consumption due to AI-controlled buildings	AI-based energy management will reduce consumption by 30% through smart buildings and predictive control. It will also enhance the use of renewable resources.

outlines estimated impacts of AI on key areas of BIM and construction up to the year 2035.

Based on the conducted analysis and predictive modeling, it can be reasonably expected that by 2035, artificial intelligence (AI) will fundamentally transform not only the BIM environment but the entire construction sector. AI will become a key element across various stages of the building lifecycle—from design and construction to management and operation. These changes will have a profound impact on the work of architects, designers, construction companies, and building managers, leading to significant digitalization, automation, and increased efficiency throughout the industry.

Significant changes are anticipated particularly in generative building design, where AI will take over a substantial portion of routine modeling tasks. This will allow architects and designers to focus more on creative and strategic aspects of design, while technical and optimization processes will be largely automated. AI will be capable of proposing optimal solutions in terms of structure, materials, energy efficiency, and costs based on defined parameters.

Another key area of transformation will be digital twins, which, when combined with AI, will become a standard tool for building management. Buildings will be operated in real time using data models that will enable not only predictive maintenance but also operational optimization and energy reduction. This technology will contribute to greater transparency, improved facility management efficiency, and enhanced building sustainability throughout their lifecycle.

Significant changes are also expected in construction and project monitoring. AI-driven autonomous drones will act as inspection tools, ensuring continuous supervision of construction quality and scheduling. Thanks to the ability to continuously collect and analyze data, these systems will identify deviations, defects, or safety risks at an early stage.

Construction robotics will lead to the partial automation of many tasks, particularly in prefabrication, assembly, and other repetitive on-site operations. The integration of AI-driven robotic systems will accelerate construction, improve accuracy and safety, and reduce the need for human labor in demanding and hazardous environments.

AI has the potential to improve energy efficiency and reduce the carbon footprint; measured outcomes will vary with data availability, local energy tariffs, and the degree of system integration. Through intelligent operations management, energy consumption forecasting, and optimization of renewable energy use, AI systems will contribute to achieving higher sustainability standards and meeting environmental goals.

These technological advancements will drive extensive digitalization and overall productivity growth in the construction sector, resulting in long-term cost savings, reduced errors, faster processes, and minimized negative environmental impacts of construction activities. AI will thus become one of the main tools driving the transformation of the construction industry towards greater efficiency, sustainability, and resilience to future challenges.

The following visualization illustrates the expected growth in the share of construction projects utilizing AI over the next decade. The development is expected to be gradual yet consistently rising, in line with technological progress, decreasing costs of AI tools, and increasing demands for efficiency and sustainability across the entire sector.

Scheme 1 presents the results of an exponential trend analysis based on available historical data, current technological capabilities, and anticipated trends in AI integration into construction practice. The projection curve clearly shows that in 2024, approximately 15% of construction projects employ AI, primarily in large infrastructure or technologically advanced projects. The gradual increase is driven mainly by the broader adoption of technologies, the development of cloud solutions, growing trust in AI tools, and increasing pressure for environmental and economic sustainability.

By 2030, scenario-based forecasts suggest that around 60–70% of projects may use AI in at least one phase of the lifecycle. Adoption will likely vary by country, company size, and discipline, reflecting differences in regulation, data maturity, and workforce skills. By 2034, AI could be integrated in a large share of projects, but not uniformly across markets.

The chart thus clearly demonstrates not only the rising importance of AI in the industry but also emphasizes the sector’s transformation towards greater digitalization, automation, and efficiency, as described in the earlier part of this chapter. For the preparation of the graph, a predictive analysis of AI adoption in construction was conducted according to the scenario presented at the beginning of this chapter. The evaluation was based on the robustness of the prediction, with outputs reported as 95% confidence intervals using bootstrap replication. These estimates should be interpreted as indicative scenarios rather than precise predictions and are sensitive to policy changes, macroeconomic cycles, and sectoral differences.

8. Discussion

Although the benefits of using artificial intelligence (AI) in Building Information Modeling (BIM) are undeniable, it is essential to openly address the challenges and risks associated with this technological transformation. A key issue remains the quality and availability of data, which serves as a fundamental input for the effective functioning of AI models. Currently, many BIM models suffer from a low degree of standardization and contain incomplete or erroneous information, which significantly limits the reliable application of AI algorithms. Without unified methodologies for data collection, classification, and storage, the full potential of AI tools—particularly in areas such as predictive analytics, generative design, or automated control—cannot be realized.

While the literature points to sizeable opportunities, real-world uptake is highly heterogeneous. (i) Between countries, regulatory maturity, procurement rules and data infrastructure strongly shape adoption trajectories. (ii) Across company sizes, large contractors and owners with dedicated CDE/IT teams scale faster, whereas SMEs face capital and skills constraints. (iii) Across professional roles, architects, engineers, contractors and facility managers adopt at different speeds and for different use cases. These differences produce distinct bottlenecks—from fragmented data governance and uneven skills to interoperability gaps and compliance risks. As a result, reported benefits should be read as conditional on data quality, standardized identifiers and change-management capacity rather than universal outcomes.

Segment-specific issues:

• Countries/regions: uncertainty in liability and procurement, limited open data → slow or selective pilots.
• SMEs: tool costs, training burden, vendor lock-in → stop-start adoption and shallow integration.
• Large organizations: cross-department interoperability and legacy systems → duplication of models, slow change control.
• Disciplines: misaligned KPIs (design vs. construction vs. FM) → value capture asymmetry and weak feedback loops.

Inference tools can inadvertently amplify bias present in training data and heuristics—e.g., over-representation of certain building types, regions, vendors or design patterns—leading to systematic errors that propagate through the common data environment (CDE). Typical failure modes include dataset shift (models applied outside the distribution they were trained on), selection bias (results reported only for successful pilots), and automation bias (over-reliance on model suggestions). To prevent error accumulation in information flows, we recommend a human-in-the-loop review, explainability (XAI) for safety-critical decisions, and a documented QA workflow (data lineage, model versioning, thresholds, and audit trails). Where possible, results should be stratified by country, company size and professional role to reveal hidden performance gaps.

Mitigation strategies. (a) Data governance first: IFC/BCF alignment, curated dictionaries and element IDs; (b) Human-in-the-loop review with traceable approvals; (c) Explainability (XAI) for safety-critical decisions; (d) Stratified roll-outs (SME-friendly bundles, shared sandboxes); (e) Procurement clauses for data portability and audit trails; (f) Capability building targeted by role (designer/PM/FM).

Importantly, adoption trajectories are not uniform. Cross-country heterogeneity (e.g., regulatory support, public procurement rules), company size (large contractors vs. SMEs), and professional specialization (designers, site managers, FM) lead to divergent capacities to implement AI. These differences can generate specific frictions—from fragmented data governance and uneven skills, to biased model performance and compliance risks. Future work should therefore report results by region, firm size, and discipline, and discuss mitigation strategies such as shared data standards, targeted training for SMEs, and explainability requirements for safety-critical use cases.

Equally serious is the economic barrier that prevents widespread adoption of AI technologies, especially among small and medium-sized enterprises. High costs for hardware, software, and specialized personnel mean that digital transformation is often the domain of large corporations. A potential solution lies in the development of the AI-as-a-Service (AIaaS) model, which allows access to tools through flexible subscription models without the need for large upfront investments. Government and EU support programs could play a key role in this respect—not only through direct funding but also by launching pilot projects, educational initiatives, and certification programs.

As AI becomes integrated into the design, construction, and operational phases of buildings, it brings with it fundamental legal and ethical questions. It is still not clearly defined who holds responsibility for errors caused by an algorithm—the designer, the software provider, or the investor? The absence of a legislative framework creates a legal vacuum, particularly risky in the context of public procurement or critical infrastructure. The ethical dimension of AI implementation also requires careful attention: decision-making algorithms often function as “black boxes,” whose outputs are difficult to verify or explain. This increases the risk of losing professional accountability, which is unacceptable in matters of public space, safety, and health.

Education of professionals will play a key role in this process. Architects, engineers, construction managers, and facility managers will need to expand their competencies to include basic knowledge of AI, data analytics, digital twins, and the interpretation of model outputs. Universities and professional training institutions should respond to this challenge by innovating curricula and introducing interdisciplinary courses that reflect the current state of technology and its impact on practice.

Looking ahead, it is reasonable to assume that AI will play an increasingly important role in all phases of the building life cycle. Already today, it assists in structural design, material flow optimization, energy performance simulations, and failure predictions. Over the next decade, AI tools are expected to become standard in both design and operational processes—not only for large projects but also for smaller, local investments. Advancements in technology, standardization of procedures, and supportive legislative frameworks will facilitate this transition.

This positive trend is also reflected in the development of scientific output, as shown in Scheme 2, which illustrates a dramatic increase in academic publications related to AI and BIM. Between 2023 and 2024, the number of new articles increased almost sevenfold. As of mid-2025, more than half the number of publications from the entire previous year has already been published, indicating growing interest from both research and application sectors, and a strong innovation drive within the construction industry.

Implications for research. Future studies should (i) report results stratified by country/region, company size and professional role; (ii) publish reproducible pipelines (data lineage, model cards, KPIs); (iii) evaluate closed-loop workflows (detection → human review → implementation → feedback) rather than isolated pilots; and (iv) include cost-to-benefit analyses for SMEs vs. large organizations.

9. Conclusions

This review indicates the potential of AI to improve efficiency, quality, and sustainability across the BIM lifecycle; however, the scale and timing of benefits are context-dependent—on data readiness, governance, regulation, and organizational capabilities. Adoption will not be uniform. Countries with supportive public procurement and clear liability frameworks, large asset owners and digitally mature contractors, and teams with well-defined data governance are likely to realize gains earlier than SMEs and resource-constrained environments. Differences across professional roles (design, construction, FM) further shape use cases, KPIs, and value capture. AI expands BIM’s capabilities in areas such as design automation, generative design, predictive analytics, data management, quality control, and operational process optimization. The article shows that this is not merely a supplement to traditional tools but a paradigmatic shift in how buildings are designed, constructed, and operated.

For practice, we recommend: (i) starting with data governance (IFC/BCF alignment, element identifiers, curated dictionaries); (ii) implementing human-in-the-loop review and explainability (XAI) for auditability in safety-critical decisions; (iii) piloting role-specific use cases with measurable KPIs and closed-loop feedback (detection → human review → implementation → monitoring); and (iv) embedding AI outputs into procurement and change-control to ensure traceability and value capture.

The synthesis of current findings indicates that the integration of data-driven and model-based AI approaches into BIM accelerates information flows between design, construction, and operation, while reducing the time between problem occurrence and resolution. The greatest benefits consistently manifest along three main lines: (i) expanding the space of tested solutions in the early design phase while maintaining controllability over criteria and standards, (ii) continuous verification of “as-designed” and “as-built” compliance with direct linkage to tasks in the CDE, and (iii) the transition from reactive maintenance to predictive operation based on digital twins. These trajectories are already being reflected in specific workflows in the domains of computer vision, generative approaches, and operational diagnostics, with direct integration into shared BIM data and quality management.

Examples of generative design, error detection, predictive maintenance, and digital twins demonstrate that the integration of AI and BIM has the potential to significantly increase efficiency, reduce costs, and enhance the sustainability of construction projects. Particularly in the operation and facility management phases, AI enables failure prediction, energy management, and time-based simulation of building performance, which constitute major innovations. It has also been shown that AI can be an effective tool in the digitalization of existing buildings through the Scan-to-BIM process and in managing complex construction projects.

However, the broader implementation of AI in the BIM environment is hindered by several barriers. The most critical among them are the economic costs of the technologies, the lack of high-quality and standardized data, software incompatibilities, and a general shortage of AI-related competencies in construction practice. A major unresolved issue remains the lack of legal and ethical regulation, which complicates matters of responsibility, transparency in decision-making processes, and data protection. AI often acts as a “black box,” producing decisions that are difficult to interpret—an issue that can be particularly risky in the construction sector.

These findings indicate that the implementation of AI in BIM should not be seen merely as a technical upgrade, but rather as a transformational change requiring interdisciplinary collaboration, shifts in professional education, new forms of accountability, and a clearly defined ethical framework. Technological innovation alone is insufficient; the construction industry also needs a cultural readiness for digital transformation.

Looking ahead, several key areas of AI integration into BIM deserve increased scientific attention. One such area is the development of explainable AI (XAI)—systems whose decision-making logic is transparent, auditable, and understandable for both professionals and the public. Transparency is especially important in the context of public infrastructure, where trust and accountability are paramount.

Another crucial direction is the standardization of data structures. The successful and efficient integration of AI heavily depends on the availability of open and interoperable data formats that enable seamless information exchange between different software tools. In this respect, the development and broader adoption of the Industry Foundation Classes (IFC) format is emerging as a strategic priority, potentially playing a key role in harmonizing data flows throughout the entire building lifecycle.

The AI-as-a-Service (AIaaS) model also offers significant potential. This approach could greatly improve access to advanced AI tools for small and medium-sized enterprises, which often cannot afford robust and costly implementations. Through cloud platforms, these companies can gain access to flexible and affordable solutions without the need for major investments in hardware or specialized teams.

As AI becomes increasingly deployed in sensitive areas of construction practice, there is a growing need to establish comprehensive ethical and legal frameworks. It is essential to define rules regarding responsibility for AI-generated outputs, intellectual property rights over algorithmically produced content, and procedures for validating AI-generated designs. These frameworks are critical for ensuring safety, quality, and public trust in emerging technologies.

Finally, the development of participatory AI systems represents an equally important area. These systems would enable the involvement of various stakeholders—from architects and designers to facility managers and end-users—in AI-driven decision-making processes. Such an approach can foster greater acceptance of new technologies, enhance process transparency, and promote the creation of designs that better reflect the actual needs and experiences of building users.

The empirical picture is supported by both the dynamics of scientific output and model-based adoption projections. The surge in publications in recent years confirms a strong innovation impulse and the rapid maturation of the topic in both academic and applied spheres, while forecasts based on a combination of publication time series and industry indicators point to a substantial expansion of AI use in projects over the next decade.

At the same time, certain conditions remain indispensable for realizing the expected effects: standardized and complete data, open interfaces and version traceability, curated thresholding and prioritization of findings, and systematic expert verification before committing changes to the model and schedule. These elements determine the transferability of pilot results into routine practice and the return on investment; likewise, economic accessibility and clearly defined legal and ethical frameworks will be decisive.

Overall, the results indicate that AI is becoming a structural component of the BIM data ecosystem wherever it is firmly anchored in open standards, connected to operational measurements, and evaluated through a set of comprehensible KPIs. In combination with the transparent predictive framework presented in this work (including confidence intervals and sensitivity scenarios), the AI–BIM environment provides a practical guide to shortening iterations, reducing rework, and stabilizing operations—without the ambition to replace professional responsibility, but with the aim of enabling earlier, better-informed, and repeatable decisions.

The main contribution of this study lies in the operationalization of AI integration within the BIM environment: we connect AI techniques with openBIM and CDE across design, construction, and operation, and formulate a closed loop from finding to validated intervention and auditable change in the model and schedule. It refines adoption forecasts through a transparent framework (logistic curves, 95% CI, sensitivity scenarios) and unifies evaluation criteria to enable benefits to be measured comparably across projects. It consolidates the data prerequisites for replicable performance (LOIN/IDS, threshold curation, human-in-the-loop) and proposes practical governance rules (XAI, evidence of versions and decisions) for safe deployment into practice.

In conclusion, AI in connection with BIM is not merely a technological trend, but a structural transformation of the entire construction sector. Its successful implementation requires not only advances in algorithms and data, but also a rethinking of collaboration, accountability, and professional education. Future research in this field should continue to focus not only on technical efficiency but also on sociotechnical dimensions—ensuring that AI serves as a supportive tool rather than an opaque arbiter of decisions. The future of AI in BIM is undoubtedly full of opportunities, but their responsible implementation will require a concerted effort of scientific, technical, and ethical expertise.

Quality assurance remains essential: without human-in-the-loop review, explainability for safety-critical decisions, and auditable data/model governance, AI-assisted workflows may inadvertently amplify bias and errors across the CDE. Looking ahead, AI is well-positioned to become an important enabler of digital transformation in many contexts, provided that sectoral heterogeneity is addressed and governance mechanisms mature. Under these conditions, expected benefits in time, cost, and energy are achievable, but not universal.