By Bob Tuman, safety consultant
I was right in the middle of helping a painting contractor client figure out why his workers’ compensation premium had jumped $50,000 when a New York Times article caught my attention.
As I read “A.I. Can’t Build A High Rise, But It Can Speed Up the Job” (https://www.nytimes.com/2023/08/15/business/artificial-intelligence-in-construction-real-estate.html?searchResultPosition=1) and saw robots roaming around a jobsite looking for inefficiencies which could adversely impact project completion, I had an idea. Could these same robots detect unsafe conditions and unsafe behavior?
Case in point: the now almost $1 million workers’ compensation claim which resulted in a $50,000 increase in my client’s premium was related to a 15’ fall and severe injury because the General Contractor’s employees had removed the leading edge guardrail and didn’t put it back.
The NY Times article focused on achieving efficiencies to speed up construction, but I wondered if A.I. could help to prevent construction-related accidents and deaths.
I contacted the robot manufacturers and software developers with “How about a safety component?”, but got crickets. I was disappointed but not deterred. I read everything I could find on the application of A.I. and vision systems to construction safety and became passionate about finding ways to implement it.
How Can A.I. have a Positive Impact on Construction Safety?
For the last 40 years I have been inspecting construction job sites citing unsafe conditions and unsafe behavior and complimenting contractors on safe field practices. However, I could only comment on negative or positive conditions and behavior at the point in time of my inspection. My hope was for Foremen and workers to internalize reasons and actions to protect themselves. This hope sometimes translated to enhanced attention to field safety practices, and it sometimes did not.
The large projects, like the one where I was asked by a subcontractor client to inspect weekly, have at least one full-time site safety officer tasked with being the eyes and ears of the General Contractor. This requires her or him to be on site throughout each day observing and, if needed, correcting unsafe conditions and unsafe behavior.
However, this was not always the case.
Before my first weekly inspection, I spent time with the General Contractor’s full-time safety officer. I wanted him to know that I was there as another set of eyes primarily focused on my client’s attention to safety but also on unsafe conditions which could adversely impact the health and safety of my client’s employees. And before going on site each week, I went to the General Contractor’s trailer to check in with the site safety officer. He routinely asked me to let him know what, if any, unsafe behavior and unsafe conditions I observed. However, my client and I decided that it was best to transmit my findings to its General Superintendent who would then relay them to the General Contractor’s Superintendent.
Regrettably, it was disconcerting to repeatedly find “IDLH” (immediate or imminent danger to life and health) unsafe conditions, and to see the General Contractor slow walk corrective actions. For example, I found unguarded elevator shafts on this 5-story building 2’-3’ from my client’s electrical rooms. Had an electrician entering or exiting these rooms accidentally slipped and fallen, he or she would have been rendered a paraplegic, quadriplegic or dead.
I reported this immediately to our General Super who immediately reported this to the GC’s Super, but the site safety officer and the GC’s Super took no corrective action. And each week for the next 6-8 weeks I observed, photographed and documented this same and other high gravity unsafe conditions. And each week I repeatedly asked our employees not to work anywhere near the unprotected elevator shafts. Finally, 9 weeks after I first observed this unsafe condition, the site safety officer and the General Super guarded the elevator shafts.
During another inspection, I observed a client’s employees working on unguarded pipe scaffolding. The scaffolding was missing guardrail, end rails and kickboard (falling object protection). I recommended that they work elsewhere until and unless this was corrected. Our General Super reported this to the GC’s Super, but again the site safety officer and GC’s Super did nothing for weeks.
WHAT WAS GOING ON?
My client’s employees told me that the site safety officer was only periodically walking the job site. The reason: this morbidly obese man could not make it up the stairs without becoming winded and fatigued. Our employees saw him perhaps once a week. Fortunately for everyone, the GC eventually replaced this guy and brought on a competent and motivated safety professional.
The above aside, it is a fact that regardless of how diligent site safety officers are, they cannot police safety every minute of every day. For example, if a worker operating a nail gun decides to remove her safety glasses right after the safety officer leaves her work area, there is nothing that will stop her from sustaining an eye injury.
So, What Can A.I. do to Minimize the Probability of Accidents with Injury or Death? How about Drones, Robots, and/or Other Vision Systems (i.e. CCTV)?
Are the stakes high? You betcha. Just look at the costs- both financial and human- of workplace accidents, per the 2022 National Safety Council “Injury Facts” (https://injuryfacts.nsc.org):
63 million occupational injuries
227,039 preventable deaths. Construction accidents accounted for 1,018 fatalities
270,000 construction medically consulted injuries
Cost of occupational injuries and deaths: $1.284 billion
Here are some other reasons to try A.I. or vision systems:
Mouth open and no harness, roofer falls into 525-degree kettle of liquid asphalt (per OSHA “Fatality Reports”)
Apprentice electrician’s arms pulled off at his shoulders while operating a malfunctioning wire puller (per Boston Globe)
Worker sits on propane tower heater, igniting his pants and private parts (per a lawsuit against the GC, the tower heater distributor, and the tower heater manufacturer for lack of consortium)
Tailgater pulls crossbow out of his trunk and shoots the other driver (per the Boston Globe)
$60,000 worth of pipe scaffolding “walks off” the jobsite, stunning the General Contractor (per a General Contractor client)
Could A.I. have prevented these fatalities and catastrophes and incidents? Maybe, maybe not. But it is worth a try.
What might a construction AI or vision system look like?
- Video cameras strategically placed throughout a jobsite filming and archiving workers and work, robots roaming jobsites, drones flying over jobsites, with trained and experienced construction safety professionals stationed at monitors, identifying and alerting construction management in real time to unsafe behavior and unsafe conditions.
To neutralize the potential of project management claiming “Why are you telling me that the sky is falling when you only saw one guy operating a nail gun without eye protection?”, the construction safety professionals would adhere to negotiated ratings’ guidelines and response timelines also agreed upon prior to the first shovel going into the ground.
What do you think of the following ratings:
- No Unsafe Behavior and/or Unsafe Condition
- Low Gravity
- Low to Medium Gravity
- Medium Gravity
- Medium to High Gravity
- High Gravity
- Immediately/Imminently Dangerous to Life or Health (“IDLH”).
Let’s play this out with some hypotheticals:
Workers taking morning break in their work area remove their hardhats. How would you rate this? My rating: No Unsafe Behavior
Workers on break in their work area without eye and head protection near stud and drywall workers operating nail guns. Your rating? My rating: Medium to High Gravity, requiring corrective action sooner than later.
Spoils piles 1’ from the edge of an unoccupied 8’ deep trench. Your rating? My rating: Low Gravity.
Spoils piles 1’ from the edge of an occupied 8’ deep trench. Soil rating: unstable (water seen coming into the trench from the sides and bottom of the trench). Your rating? My rating: High Gravity. Get the workers out of the trench right away and install cave-in protection.
Roofers without fall protection working 1’ from the roof’s edge. Your rating? My rating: Imminent Danger to Life and Health, requiring immediate corrective action.
Roofers working 20’ from the edge. No roof openings. Your rating? My rating: No unsafe behavior or unsafe condition.
- Drones, robots and CCTV systems programmed to identify actual or potential safety hazards, alert project management to take corrective actions commensurate with the magnitude of the hazards, verify that the actual or potential safety hazard(s) has been corrected, and archive corrective actions.
This is a more complex and challenging effort, but not insurmountable. This would require capturing images, image recognition and image comparison.
For example, software developers would store images of construction workers wearing Personal Protective Equipment engaged in various tasks- i.e. operating nail guns, hand digging excavations, erecting steel, installing studs and drywall, or coring or grinding concrete. The robots, drones or CCTV cameras would then compare real-time images of workers to determine compliance or lack thereof with required exposure-specific Personal Protective Equipment (“PPE”), send these images to project management who would judge the image and the need for corrective action, per the rating system in number 1 above. If merited, project management would then and initiate corrective action.
What is the downside of number 2?
- Information overload
- Garbage in, garbage out
- False positives and “jump the gun alerts”. Example: I’ve run across employees who removed their safety glasses for seconds to clean them or wipe the sweat from their eyes and face. Does this rise to the level of the need for an alert? I don’t think so.
- Resistance from project developers and General Contractors who perceive that enhanced scrutiny of workers’ attention to safety will slow things down and impede production.
Maybe, but the opposite could be true. Let me tell you why. A local large project was completely shut down for 3 weeks by the state fire marshal, setting the already late project back months. Had a robot identified fire and life safety hazards AND the General Contractor attended to fire and life safety, this never would have happened.
Subcontractor push back. This can be diffused by giving them a say in the rating development and implementation process.
What is the upside of AI?
Increased compliance with safety rules and generally accepted safe work practices. If you knew you were on “candid camera”, would you take your hard hat off when, for example, other contractors were working above you, with construction debris falling on you?
Identifying and mitigating construction inefficiencies.
Minimizing sabotage and theft. A General Contractor had to pony up $60,000 when rented pipe scaffolding “walked off” the jobsite over the course of several weeks.
Creating a permanent electronic record to minimize current and future liability exposure. Large projects can generate aggravating and costly lawsuits. For example, a school building contractor was sued when a pedestrian claimed she slipped and fell on an uneven sidewalk. With the help of a CCTV camera on a building adjacent to the job site, the investigation found that she had faked the fall.
Lower insurance costs. A great many projects are OCIP (Owner controlled insurance program) or CCIP (Contractor controlled insurance program) with high deductibles.
Workers feel safe. I can state unequivocally that projects where safety is a priority and where deviations from generally accepted safe work practices are addressed immediately generate productive work practices and come in under budget. If you were a worker bee, wouldn’t you feel safer and be more inclined to comply with your and the GC’s safety guidelines if there was scrutiny throughout the day?
Who is Going to Pay for A.I., Robots, Drones, or CCTV?
The answer is: Who benefits from exemplary loss experience?
- Workers’ compensation insurers. How about incentivizing contractors which implement AI with credits and dividends?
- General contractors and subcontractors which get credits and dividends for favorable loss experience.
- General contractors and subcontractors which come in under budget.
In closing, how about giving it a try? Maybe number 1 is a way to dip your toes in the water. Then maybe you will want to take a leap of faith into number 2.
About the Author:
Bob Tuman is president of CCR Safety Consulting in California, providing safety consultation to construction contractors and performing Workers’ Compensation and General Liability Loss Control Surveys for property and casualty insurers. For further information or to contact CCR directly, please contact: 805-545-5976 or
email bobtuman@gmail. com
