How AI could help make the workplace safer

(Note: This is the first in a three-part series on the use of artificial intelligence tools in the workplace)

There’s a lot of concern about the potential risks or downsides of artificial intelligence — a term that encompasses technologies such as machine learning, computer vision, and predictive analytics — but it has its positive aspects as well. For one thing, it is rapidly expanding the capabilities of safety leaders in modern workplaces, and represents a profound departure from traditional safety measures, which often rely on reactive responses to incidents after they have occurred.

Instead, AI allows for a proactive approach, giving companies the ability to anticipate and mitigate hazards before they escalate into accidents by using tools such as predictive analytics, real-time monitoring, and automated hazard detection. Doing so can foster a safer work environment, and substantially reduce accident rates — and premiums.

Machine learning algorithms are instrumental in industrial risk assessment, discerning complex patterns, correlations, and trends from massive amounts of data to forecast potential hazards. This process relies on comprehensive data collection from diverse sources, including historical incident reports, near-miss data, safety inspections, employee observations, equipment logs, and real-time sensor readings.

For instance, AI models can analyze historical incidents and near-miss patterns to predict where future risks are likely to emerge, enabling the implementation of highly targeted preventive actions. AI can also generate customized checklists to ensure adherence to occupational safety regulations.   

Unlike traditional CCTV systems that require manual oversight, cameras with AI built in or added downstream can actively analyze video footage to detect risks instantaneously, triggering immediate alerts. This capability stems from AI’s ability to process high-resolution, real-time visual and sensor data at a granular level, far exceeding human capacity for continuous observation.

Fisher Phillips, a US law firm that specializes in workplace safety cases, says that one of the most impactful ways AI can improve workplace safety is by enabling real-time hazard detection. Traditional safety measures often rely on periodic inspections, manual monitoring, and reactive responses to incidents. AI, on the other hand, could allow for continuous monitoring of environments using sensors, cameras, and wearables.

Tracking potential safety hazards in real time

In a number of workplaces, including factories, warehouses and other facilities, AI-powered computer vision systems are analyzing live video footage to detect unsafe practices, such as improper handling of hazardous materials, workers not wearing personal protective equipment (PPE), or unsafe machinery operations. These systems then alert supervisors or workers in real time, preventing accidents before they occur.

In fact, OSHA inspectors are planning to use AI-powered smart glasses equipped with cameras, sensors, and internet connectivity for real-time documentation, augmented reality features, and instant communication with outside personnel. 

AI’s predictive capabilities represent another game-changing benefit for workplace safety professionals. Machine learning algorithms can analyze vast amounts of historical data, including accident reports, near-miss incidents, and environmental factors, to predict potential hazards before they happen. Entrepreneur Philip Lorenzo has developed software that creates daily digital models of work progress from videos and images, known as “reality capture.”

The tool, called SafetyAI, analyzes each day’s reality-capture imagery and flags conditions that violate Occupational Safety and Health Administration rules, with what Lorenzo claims is 95% accuracy. Launched in October 2024, it’s now being deployed on hundreds of construction sites in the US, as well as in locations in Canada, the UK, South Korea, and Australia.

SafetyAI is one of a number of construction safety tools that have emerged in recent years. Many rely on teams of human “clickers” to manually draw boxes around images of key objects like ladders, in order to label large volumes of data and train an algorithm. SafetyAI uses an algorithm to recognize objects such as ladders or hard hats and then uses software to “reason” about what is going on in an image of a site and draw a conclusion about whether there is a safety violation.

Using years of reality capture imagery gathered from customers, with their explicit permission, Lorenzo’s team has assembled a data set of tens of thousands of images of OSHA violations. But as the 95% success rate suggests, Safety AI is not a flawless and all-knowing intelligence. It requires an experienced safety inspector as an overseer.

In the second part of our series on AI and workplace safety, we will look at some of the practical applications of the technology as well as some of the ethical and other challenges involved.

Interested in learning more about what a WCB employer representative can do for your group? You can connect with us on Facebook, Twitter, or Linkedin. Or you can reach us via email at [email protected], or you can contact us directly, during business hours, using our chat feature or by telephone at 1-844-377-9545.