Recently, discussions about the transformative power of Artificial Intelligence (AI) have become ubiquitous, extending from technological forums to casual conversations. Indeed, AI has already begun to reshape numerous aspects of our lives, and there is a general consensus that we’ve only begun to scratch the surface of its potential applications. For those of us dedicated to enhancing construction safety and ensuring crew well-being, the pivotal question arises: How might AI be leveraged to enhance our mission?

Before delving deeply into discussions about AI, it’s crucial to establish clarity, particularly regarding what AI is and is not. The prevalent discourse about AI often muddles its distinction from other technologies such as Machine Learning and predictive analysis. While there is no denying the substantial capabilities of these technologies, my aim here is to maintain a strict focus on AI, ensuring our dialogue remains undistracted and pointed.

According to Google, “Artificial Intelligence is a broad field, referring to the use of technologies to create machines and computers capable of mimicking cognitive functions associated with human intelligence, such as recognizing and responding to spoken or written language, analyzing data, making recommendations, and more.” Bearing this definition in mind, particularly considering the aspect of mimicking our cognitive functions (minus its limitations), where can AI provide crucial assistance?

In this article, I am concentrating on three essential safety functions that significantly depend on our cognitive abilities and explore how AI can be utilized to enhance these functions.

Toolbox Talks

Toolbox talks are brief, informal meetings focused on safety topics related to specific jobs. Conducted on job sites and lasting typically between ten to fifteen minutes, they aim to refresh workers’ knowledge of safe practices, heighten awareness of potential job-specific hazards, and facilitate discussions about preventing accidents. Led by a supervisor or safety expert, these talks encourage workers to share experiences and insights, promoting a collaborative approach to maintaining workplace safety.

Regulations frequently mandate in-person safety training, such as toolbox talks. While AI might not directly engage in training construction crews, its capabilities, including understanding spoken or written language and making recommendations, position it ideally to bolster safety training. AI can supplement training material and complement the trainer’s role.

Enhancing safety toolbox talks with real-time assistance and analysis

During safety toolbox talks, a critical aspect of workplace training, the host might occasionally overlook essential points or find themselves without access to the most contextually appropriate training materials. Leveraging the capabilities of modern AI systems can provide a solution to these challenges. Such systems can offer real-time transcription of the session and a comprehensive analysis of the content presented. In doing so, they can assist the trainer by pinpointing any inadvertently missed topics. Moreover, these AI tools can suggest and provide supplemental material that aligns with the context of the job, ensuring that the foreman and the team have a complete and thorough understanding of the safety procedures needed for the jobs.

Post-training assessment and reinforcement

Quizzes conducted after toolbox talks not only serve as an effective tool for gauging learning outcomes but also act as a reinforcement mechanism to solidify the acquired knowledge. One inherent challenge in this process is the significant effort required to craft apt questionnaires. This becomes even more daunting when the training content is dynamic and ever-evolving. Fortunately, advancements in AI can provide a solution. Modern AI systems can meticulously analyze the content delivered during the training and generate tailored quizzes based on it. This empowers trainees to immediately test and reinforce their understanding right after the training session, thus enhancing the overall efficacy of the learning process.

Optimized micro-training  

Micro-training stands out as a particularly effective method for training crew members. Implemented almost daily, these concise training sessions present crew members with essential information or pose a singular, focused question to them. This approach not only reinforces knowledge but also ensures consistency in learning, especially when conducted after their daily sign-ins. In the realm of technology, generative AI systems can be a game-changer in this context. They can craft specialized micro-training content, taking into consideration the broader training materials. Furthermore, each session can be meticulously tailored to individual crew members, reflecting their unique training history, past performance, and the specific nature of tasks they’re poised to tackle, ensuring a more personalized and effective learning experience.

Hazard Analysis

JHA (Job Hazard Analysis) and THA (Task Hazard Analysis) are techniques designed to pinpoint and assess hazards tied to specific jobs or tasks. Their primary aim is to streamline tasks in a manner that diminishes or eradicates potential risks to workers. This involves dissecting a job into distinct steps, identifying hazards at each stage, devising preventive strategies, and communicating them to the crew.  

By employing AI’s capabilities, JHA/THA can be analyzed in real time, ensuring that every critical component is given due attention. This real-time feedback mechanism not only ensures adherence to safety protocols but also enhances the quality of oversight. Furthermore, with advancements in generative AI models, there’s an added advantage: these systems can proactively assist foremen in detecting potential safety risks. Beyond mere identification, they can also propose well-informed mitigation strategies, making the entire safety process more comprehensive and proactive.

AI may also help to enhance and assess the crew’s understanding of job or task-related safety risks and mitigation techniques. Employing strategies similar to those used in safety training, AI may be used to craft questionnaires and micro-training materials to gauge and reinforce the crew’s grasp of potential risks.

Ultimately, AI can be pivotal for safety managers in evaluating JHA/THA submissions from crews. This ensures thoroughness and allows for the proactive detection of potential hazards or overlooked mitigation approaches.

Safety Observations

In the realm of enhancing safety through observation, human intelligence remains unparalleled—if only we could harness its full potential. Crew members, dispersed across job sites with their keen eyes and ears, possess an innate ability to detect hazards, an aptitude that AI has yet to match. Yet, possessing such capabilities and actively using them to identify and report hazards are distinct matters. The strength of human discernment is, at times, curtailed by our inherent limitations. This is precisely where AI can provide invaluable support. 

Indeed, AI is already enhancing safety in various ways, such as detecting fatigue through facial expression analysis and identifying physical hazards by evaluating live job site videos. However, it’s in the realm of mental health and suicide prevention that AI holds the potential to truly revolutionize our approach to make a significant impact.

According to the Centers for Disease Control and Prevention (CDC), construction and extraction occupations have the highest proportion of suicides among males. Notably, the suicide rate in the construction sector is nearly double that of the general population. In 2020, close to 20% of adults in the US were grappling with a mental illness. Consequently, issues of mental health and suicide significantly impact society at large and particularly pervade the construction industry.

This area also highlights the intrinsic limitations of our observations or actions on those observations. Frequently, signs of mental health issues among crew members go unnoticed. Even when detected, they often are not addressed due to the prevailing stigma surrounding mental health. This is precisely where AI harbors the potential to make a substantial difference.

AI offers promising avenues for augmenting mental health strategies, especially within high-stress environments like the construction industry. Utilizing advanced algorithms, AI can analyze various behavioral, vocal, and facial cues to detect subtle, often overlooked signs of mental strain or distress among crew members. Not only does this allow for the early identification of individuals who may be grappling with mental health issues, but it also fosters a proactive approach to addressing these challenges. AI-driven platforms can discreetly direct affected individuals towards appropriate resources, support channels, or professional help, thereby mitigating the impact of stigma associated with openly discussing mental health in the workplace. Furthermore, integrating AI ensures a consistent, non-judgmental, and unbiased monitoring system that enables organizations to sustainably support their workforce’s mental well-being, ultimately contributing to enhanced overall safety and productivity on the jobsite.

These are merely a few examples of how AI can be utilized to enhance job site safety in the industry, and undoubtedly, these instances are just the tip of the iceberg. However, regardless of whether these or other possibilities emerge, there are at least two key challenges that warrant our attention. Firstly, the effectiveness of any AI solution hinges on the quality and relevance of the information used during its training phase; thus, the training process of such AI is crucial. Secondly, and potentially more perilous, is the risk of complacency. Knowing that an AI system is vigilant, will we lower our guard? Unlike many situations where AI serves as a convenient tool, in safety, it also becomes a tool of consequence, necessitating careful consideration and strategic implementation.