Workplace hazards have long plagued companies, their employees and their profitability. Safety managers and teams can benefit from an AI-driven digital transformation, much like their counterparts in finance and marketing departments.
The U.S. Bureau of Labor Statistics recorded 5,190 fatal work injuries in 2021, the most recent year for which data is available. Additionally, there are many more non-fatal accidents in dangerous environments. The human loss and pain is unnecessary and avoidable, while the business costs are manifold. Clearly, industrial safety should be a high priority for businesses.
AI tools and techniques are applicable in many safety use cases. AI-enabled safety technologies include drones, robotics, wearables, sensors, smart equipment, augmented reality (AR) and virtual reality (VR), computer vision, mobile apps and analytics software. The hardware components collect data (e.g., sensors) or execute actions (e.g., robots). Software components rely on machine learning to analyze patterns and generate insights on safety hazards.
Business leaders should assess the breadth of uses for AI technology in industrial safety. The culture and risk management policies within their businesses should also be conducive to adopting AI tech for safety purposes.
Applications of AI for industrial safety
According to the National Institute for Occupational Safety and Health in the U.S., the fundamental method to protect workers is to control their exposure to occupational hazards. The NIOSH hierarchy of controls starts with personal protective equipment (PPE). Next are organizational controls that influence how employees work, and engineering controls that isolate employees from a hazard. Finally, there are methods of substitution (replacing the hazard) and elimination (removing the hazard altogether).
Artificial intelligence can help at each of these levels in the following ways:
- Safety rules compliance and hazard identification. Workers are not always 100% compliant with PPE requirements because they find the equipment too cumbersome or take safety for granted. A computer vision solution, for example a CCTV system with AI-enabled cameras and software, can monitor designated workplace areas for PPE noncompliance.
- Identifying hazardous objects. In this case, hazardous objects include debris and spills that can cause injuries. Over time, safety patterns and trends can be discerned and linked to enterprise data and used to improve safety and operating performance.
- Monitoring for fatigue symptoms. Workers operating dangerous and heavy equipment or vehicles need to be alert at all times. Facial expression analysis, on-site or in-vehicle, can identify signs of fatigue or drowsiness. The employee can receive an alert and be advised to resume work after a period of rest.
- Fall detection in construction. Many worker injuries in construction are due to falls. AI-powered fall detection software, often in a form as simple as a phone app, is designed for timely detection.
- Site inspections using drones. Drones and autonomous vehicles can be used to monitor and inspect construction and other hazardous sites, instead of putting employees at risk.
- Conversational AI for safety. Chatbots trained on safety procedures and manuals can answer employees' safety-related questions using natural language processing.
- Incident reporting using voice. Using voice makes it easy for employees to report incidents. AI can transcribe spoken incident reports and extract relevant data for further analysis.
- AR for equipment repair. Using AR, employees receive maintenance instructions with real-time information about diagnosis and repair.
- Safety trainings in VR. Safety training is a good fit for VR, as it simulates different hazardous scenarios. Employees can practice their responses in a controlled environment.
An enterprise roadmap for industrial safety
AI and smart technologies are only one piece of the puzzle. Organizational culture and operational risk management processes are equally important.
There are both tangible and intangible costs of workplace accidents. Monetary costs include, in the U.S., Occupational Safety and Health Administration, commonly known as OSHA, violation fines; healthcare expenses for employee treatments; worker compensation claim payouts and legal costs. Less quantifiable costs include productivity losses due to medical leaves, decreases in employee morale and damage to brand reputation. Therefore, the case is strong for AI adoption in business safety scenarios. The status quo is unsustainable, and AI helps mitigate these problems.
AI is instrumental when a business needs to improve its data practices. Data-driven approaches to workplace safety are more likely to be adopted when other parts of an organization are also fostering a data-driven culture. Many organizations don't collect -- and might not have mechanisms in place to collect -- the operational and process data required to ensure worker safety. Even when data is collected, they are used for reactive reporting, not for root cause analysis or proactive improvements.
Enterprise safety and risk managers must assess their current safety practices and explore opportunities to shift away from manual, compliance-oriented approaches to data-driven, proactive ones.