Sign Up
ENعربي
All articles

Project Management

Artificial Intelligence Revolutionizes Construction

  • Introduction

The construction industry is a critical component of the U.S. economy, employing over 8 million people and contributing nearly $7 trillion annually to built environment and infrastructure projects. However, the sector has been slow in adopting digital innovations and artificial intelligence (AI) compared to industries like transportation, manufacturing, healthcare, and aviation. This lag presents an opportunity for significant advancements in cost reduction, risk management, and productivity through the strategic implementation of AI technologies. This literature review examines the intersection of AI in construction with a focus on human factors, emphasizing the importance of understanding the impact of AI on construction workers' experience, safety, performance, and health. The primary objective is to summarize, synthesize, and critically evaluate existing research to provide insights into the current state of human-AI partnerships in the construction industry.

  • Background: AI in the Construction Industry

AI offers the potential to revolutionize the construction industry by addressing persistent challenges such as inefficiency, safety hazards, and workforce shortages. Unlike other sectors where AI has seen widespread adoption, construction is just beginning to explore the possibilities of AI-powered tools and technologies. Interest is growing in leveraging AI for various applications, including data-driven design and planning, automation, and real-time monitoring. AI-based systems can analyze historical data to optimize workflows and enhance on-site productivity, marking a significant step forward for an industry that has seen only slight improvements in productivity over the past few decades. However, most existing AI-related research and prototypes focus on narrowly defined problems in pre-planning/scheduling, construction safety, and productivity, often overlooking the critical aspects of such implementations on the workers themselves. This review addresses this gap by focusing on the impact of AI on human workers, an area that is currently under investigation in other domains under the general theme of human-AI interaction.

  • Thematic Areas of AI Application in Construction The application of AI in construction, as it relates to human factors, can be grouped into two main categories: (1) workers’ safety, performance, and productivity and (2) workers’ health.
  • Workers' Safety, Performance, and Productivity: AI applications in this category aim to eliminate physical collisions, improve safety through equipment and worker activity detection, and enhance overall productivity. Studies have explored the use of AI to monitor scaffolding structures, predict worker trajectories, and ensure the use of personal protective equipment (PPE). For example, Cho et al. (2018) and Sakhakarmi et al. (2019) used support vector machines (SVM) and strain data to detect scaffolding structural failures. Wang et al. (2019), Cai et al. (2020), and Siddula et al. (2016) utilized construction photos to measure risk and safety performance, with Wang et al. (2019) employing crowdsourced labeled data to detect complex construction scenes and enable vision-based workplace safety. Cai et al. (2020) used sequence-to-sequence data along with a long short-term memory (LSTM) model and wearable sensors to predict workers’ trajectories multiple steps ahead. Nath et al. (20…

AI also plays a role in equipment activity recognition, which impacts both safety and productivity. Golparvar-Fard et al. (2013) used video data with an SVM model to achieve activity recognition up to 98.33%, while Akhavian and Behzadan (2015) used smartphone-based sensors and radio-frequency identification (RFID) smart tags along with an artificial neural network (ANN) model to achieve 98.59% accuracy. These studies suggest novel methods to detect performance through activity detection and take corrective actions. Akhavian and Behzadan (2016), Kim and Cho (2020), and Ogunseiju et al. (2021) used wearable devices (i.e., smartphones, motion sensors, IMU) along with AI to detect workers’ activities, achieving more than 90% accuracy for activity prediction using DL algorithms (i.e., ANN, LSTM, CNN).

  • Workers' Mental and Physical Health: This area focuses on using AI to detect fatigue, ergonomic risks, and stress among construction workers, as well as preventing work-related musculoskeletal disorders (WMSDs). Studies have utilized wearable sensors (e.g., EEG, infrared temperature) to estimate workers’ fatigue, with Aryal et al. (2017) achieving up to 82.6% accuracy in predicting fatigue using a boosted tree classifier. Nath et al. (2018), Akanmu et al. (2020), Zhao and Obonyo (2020), and Mudiyanselage et al. (2021) used wearable sensors (i.e., smartphone, IMU, EMG) to detect awkward and unsafe body postures that might cause WMSDs. Nath et al. (2018), Zhao and Obonyo (2020), and Mudiyanselage et al. (2021) achieved 90.2% accuracy, 0.911 F1 score, and 99.35% accuracy, respectively. Additionally, Jebelli et al. (2018) used EEG signals with an SVM model to detect occupational stress with an accuracy of 80.32%, later improving the accuracy to 84.48% by using wrist wearable biosens…

AI has also been used to predict physical demands and workers' inattentiveness. Jebelli et al. (2019) and Tang and Golparvar-Fard (2021) detected physical demand using a combination of different technologies. Tang and Golparvar-Fard (2021) used photos and video data with a relatively more complex DL model to achieve 86.6% accuracy, while Jebelli et al. (2019) used data from wrist wearable biosensors and an SVM model to achieve 90% accuracy. Kim et al. (2021) and Lee et al. (2021) coupled SVM with wearable sensors to detect workers’ inattentiveness and perceived risk, respectively.

  • Human-AI Interaction in Construction

Collaboration between humans and AI is essential to leverage the strengths of both. Humans bring creativity, ethical considerations, and visionary thinking, while AI excels at extensive data analysis and quick access to information. Successful human-AI collaboration requires defined tasks and responsibilities, as well as intensive interaction. Stable underlying data structures are needed for both humans and AI. Furthermore, establishing and calibrating trust between human workers and AI systems is critical, with factors like system interface, functionality, level of automation, and explainability playing key roles. The "Five M's framework" (Man, Machine, Material, Method, and Management) highlights that any given field operation/task is not merely a collaboration between humans and machines but is also affected by organizational management.

  • Methodological Approaches in the Reviewed Literature

The reviewed literature employs various data collection methods, including wearable sensors, cameras, drones, and field sensors. Wearable sensors, such as accelerometers, gyroscopes, and electromyography (EMG) sensors, are used to capture data on workers' movements, physiological signals, and posture. Cameras and drones are used for visual data collection, enabling the monitoring of site activities, equipment operation, and safety compliance. Field sensors, such as strain sensors, are used to monitor the structural integrity of scaffolding and other temporary structures.

AI techniques and models used in these studies include SVM, CNN, LSTM, ANN, and others. SVM is used for classification tasks, such as detecting scaffolding failures and recognizing construction equipment activities. CNN is used for image-based analysis, such as detecting PPE and recognizing workers' actions. LSTM is used for sequence prediction, such as predicting worker trajectories and recognizing activities over time. ANN is used for a variety of tasks, including activity recognition, performance prediction, and gesture recognition. The accuracy and performance of these AI models vary depending on the application and the quality of the data, with some studies reporting accuracy rates above 90%.

  • Points of Agreement, Debate, and Gaps in the Research

The literature shows consensus on the potential of AI to improve safety and productivity in construction. AI-based systems can monitor worker behavior, detect hazards, and provide real-time feedback to prevent accidents and injuries. However, debates exist regarding the effectiveness of AI in specific applications and the impact on worker well-being. Some studies suggest that AI can reduce physical demands and improve worker comfort, while others raise concerns about the potential for increased stress and mental health issues due to constant monitoring and automation.

Gaps in the research include a limited understanding of the long-term effects of AI on construction workers' health and the need for more research on mental health issues. There is also a need for more comprehensive studies that consider the social, ethical, and organizational implications of AI implementation in construction. Further research is needed to develop human-centered AI systems that are transparent, explainable, and trustworthy.

  • Development in the Field Over Time

The research focus has evolved from initial studies on basic safety applications to more recent investigations into worker health and human-AI interaction. Early studies focused on using AI to detect and prevent accidents, such as falls from heights and collisions with equipment. More recent studies have explored the use of AI to monitor worker fatigue, stress, and ergonomic risks. There is an increasing use of advanced AI techniques, such as deep learning, and the integration of various data sources for comprehensive monitoring. The field is moving towards a more holistic approach that considers both the technological and human aspects of AI in construction.

  • Critical Insights and Contextualization

The reviewed body of work aligns with research in other industries regarding human-AI collaboration, but also diverges in context due to the unique characteristics of the construction industry. The dynamic and fragmented nature of construction sites, as well as the physically demanding tasks and harsh environments, present unique challenges for AI implementation. Compared to more structured and controlled environments, such as manufacturing plants, construction sites are constantly changing, making it difficult to develop AI systems that are robust and reliable.

Opportunities for AI implementation in the construction industry include the development of personalized safety training programs, the optimization of work schedules to reduce fatigue, and the creation of collaborative robots that can assist workers with physically demanding tasks. Future research should focus on developing AI systems that are adaptable to the changing conditions of construction sites and that can effectively communicate with and support human workers. A holistic approach is needed that considers both the technological and human aspects of AI in construction, with a focus on promoting worker well-being and creating a positive technology experience.

  • Conclusion

AI has the potential to transform the construction industry by improving worker safety, productivity, and well-being. However, successful implementation requires a human-centered approach that considers the impact of AI on construction workers. Continued research and collaboration are needed to fully realize the potential of AI in construction while prioritizing the well-being of its workforce. By addressing the challenges and embracing the opportunities of human-AI partnerships, the construction industry can create a safer, more efficient, and more sustainable future.

FAQ: Human-AI Partnership in the Construction Industry

  • Why is AI adoption in the construction industry lagging compared to other sectors like manufacturing or transportation?

While construction significantly contributes to the economy, its adoption of digital innovations and AI is still in its early stages. This is attributed to the industry's unique characteristics, such as its dynamic nature, fragmented workflows, and physically demanding tasks. Surveys also suggest that construction is the second least digitized global industry after agriculture and hunting. However, this presents a significant opportunity for cost reduction, risk management, and productivity improvement through AI implementation.

  • How can AI improve safety on construction sites?

AI applications focused on worker safety include real-time detection of unsafe conditions, activities, or equipment, prediction of potential accidents, and monitoring of workers' physical state. These are achieved using technologies like wearable sensors (measuring fatigue or stress), cameras (detecting PPE non-compliance or hazardous activities), drones, and computer vision. The goal is to provide early warnings and interventions, ultimately reducing the risk of accidents and injuries.

  • What specific technologies and AI models are being used to improve worker safety, performance, and productivity in construction?

Various technologies and AI models are being employed, including:

  • Wearable sensors: Used with machine learning (ML) models like Support Vector Machines (SVM) or Artificial Neural Networks (ANN) to detect fatigue, awkward postures, and stress levels.
  • Cameras and Computer Vision: Used with Convolutional Neural Networks (CNNs) to detect PPE compliance, identify hazardous activities, and track worker and equipment movement.
  • Radio-Frequency Identification (RFID) smart tags: Used to track equipment activity.
  • Drones: Used to monitor site conditions and worker activity.
  • Deep Learning (DL) algorithms: ANN, LSTM, CNN for activity prediction.
  • How can AI contribute to improving the mental and physical health of construction workers?

AI can play a role in preventing work-related musculoskeletal disorders (WMSDs) by monitoring workers' postures and movements using wearable sensors and providing real-time feedback. AI systems can also detect fatigue and stress levels through physiological measurements, enabling timely interventions to prevent burnout and other health issues. AI-powered VR training systems can help enhance the understanding of risks associated with unsafe body posture.

  • What are the benefits of combining human capabilities with AI in construction?

Humans excel at tasks requiring creativity, ethical considerations, and visionary thinking, whereas AI is adept at extensive data analysis and quick access to information. A human-AI partnership can leverage the strengths of both, resulting in solutions that are not only efficient but also ethical, inclusive, and adaptable to the dynamic nature of construction projects. This collaboration ensures that technological advancements complement rather than replace human skills.

  • What are some potential barriers to the successful adoption of AI in the construction industry?

Key barriers include a lack of understanding of how AI affects human workers, a lack of complete and accessible information about AI and BIM technologies, and the need to establish trust in AI systems. Trust is built through explainability, reliability, robustness, and demonstrated safety of the integrated technology. Effective methods of establishing and calibrating trust between humans and AI are needed, such as system interface, functionality, level of automation, and explainability.

  • How can we ensure a positive experience for workers as AI is integrated into construction processes?

Focusing on usability, interpretability, and efficacy for the user are very important. AI should be implemented as a helping hand, not a replacement for human workers, ensuring that it enhances their capabilities and improves their work experience. A positive technology experience will motivate workers to adopt AI spontaneously. It is also important to have defined tasks and responsibilities.

  • What are some future research directions for human-AI partnership in construction?

Future research should focus on understanding and promoting workers' health, addressing both physical and mental health issues related to excessive workload and stressful work environments. Research should also concentrate on job site ergonomics, safety, and performance, recommending operational and policy changes to eliminate contributing factors to health and safety problems in construction. Furthermore, studies should be conducted to improve appropriate workforce training

(Sakib & Behzadan, 2025)

Reference:

Sakib, N., & Behzadan, A. H. (2025). Human-AI Partnership to Improve Construction Workers’ Experience on Safety, Performance, and Health: A Systematic Review of The North American Construction Industry. In Journal of Engineering, Project, and Production Management (Vol. 15, Issue 1). Engineering Project and Production Management. https://doi.org/10.32738/JEPPM-2025-0006