October 17, 2023 How AI is Battling Decades of Bias in Safety Protocols

Manufacturing workforce diversity has come a long way, safety protocols have not – until now, enter AI.

By Heather Chapman, MS, CSP, CHMM, CEAS III – Head of Ergonomics, Soter Analytics

A noted study published in the American Journal of Epidemiology found that women have a higher risk of injury in a heavy manufacturing environment. Between the study’s publishing in 2009 and the latest Labor Bureau data for the sector in 2021, there has been a 26% increase in women’s representation in manufacturing, to an estimated 4.3 million women working in the field today. 

This study isn’t the only one of its kind, more recent data has confirmed similar findings that women in the industry are subject to more occupational injuries, and at times are more prone to leaving. However, few have considered the role long-held protocols and practices for safety in these environments have failed to support the increasing diversity of the workforce.  

Where safety bias lurks in industrial workplaces

A sizable portion of safety protocol today, from safe lifting benchmarks to ergonomic recommendations, are based on a one-size-fits-all approach calling back to the days when men filled the near entirety of these roles. In order to provide recommendations most appropriate for the largest portion of workers, safety protocols were designed with “the average male” in mind.

The issue with this outdated approach is that whether you’re a woman, an older individual, or anyone who doesn’t fit the bill of “the average male” physiology, safety benchmarks like “this weight is suitable for lifting for 75% of males” are neither helpful nor safe. 

Examples of safety bias in industry don’t end at these benchmarks and processes; it’s also noticeable in more direct ways, like the availability and variety of critical safety gear.

The majority of the protective equipment on the market today, from hard hats to protective clothing, is also designed with the average male proportions in mind. According to a recent study of 174 tradeswomen conducted by The Center for Construction Research and Training, 77% said they had been exposed to a hazard unnecessarily because of unsafe safety policies and implementations, such as ill-fitting personal protective equipment. 

How AI is redefining industry’s approach to safety

The one-size-fits-all approach wasn’t designed to intentionally sideline the women who would eventually come to contribute meaningfully to sectors like manufacturing. At the time, providing safety guidelines that applied to the greatest proportion of workers made sense. Luckily, today we have technologies that can independently prescribe safety recommendations that are unique to each worker’s physiology.

AI is being used in all kinds of tools to improve safety, including wearables and computer vision solutions. No matter the vehicle, all of these tools have one common requirement to be able to dispel individualized safety insights, and that’s data.

Mechanisms like wearables or smart video monitoring systems capture millions of datapoints recording how an individual worker moves, what their ergonomic tendencies and unique risks are, and from there, can independently recommend changes to reduce injury potential.   

What’s remarkable about this approach is that it bypasses grouping individuals in any way or making assumptions about how one’s body works based on their gender, age or other characteristic. No two people are alike, and safety would never be best prescribed to apply to all women workers, or all workers of a certain age or physical condition.

When fed data consistently, these new AI tools can even pick up on more nuanced trends such as someone working through a cold or flu that might be impairing their ability to exert at their usual pace, or a new, temporary injury that has them adjusting their movements in a risky way to compensate.

The point is, injuries and the risks that lead to them are unique to each person, so preventing them most effectively requires the same approach.

As solutions like wearables and computer vision continue to gather more data as adoption increases, these datasets will be able to inform larger organizational decisions like what machinery should be automated to maximize production while reducing human worker injuries. Similarly, insurers can use this data to offer more accurate and cost-effective claims packages.

AI is the key to a future where injuries are no longer a cost of doing business and where safety is equal for all.

About the Author
Heather Chapman is Head of Ergonomics at Soter Analytics and Principal of Paradigm Safety. She is a Certified Safety Professional (CSP), Certified Hazardous Materials Manager (CHMM), and Certified Ergonomics Assessment Specialist (CEAS) with over 20 years of experience working in logistics, retail, manufacturing, academia, and hospitals. She uses practical problem solving with unique approaches to help get safety cultures back on track or energized to the next level.

Subscribe to Industry Today