Robotics in the Workplace

Helping Hands?

Robotics will forever change the landscape of the U.S. workplace. That may create new liability concerns and eliminate others.
By: | October 15, 2014 • 10 min read

It’s probably way too soon to start dreaming up insurance products that will respond to the risk that robots are going to rise up and annihilate humankind. And good luck finding the market capacity for it anyway.

However, robots and robotics are fast becoming a fixture of our reality, and the industry is poised for rapid expansion.

Advertisement




Robots, in some form or another, have been present in the manufacturing sector since 1962, when a New Jersey General Motors factory began using a robot to do spot welding and extract die castings. By the ’70s, increasingly sophisticated machines, operated by minicomputers, were being widely used for small parts assembly.

Robotics have long since moved away from the assembly line. Robots are present everywhere from warehouses to hospitals, from farms to laboratories, and from the military to mines and more.

One of the latest robotic forays into the workplace is at the Aloft hotel in Cupertino, Calif., where “The Botlr” — a service robot that looks like a distant cousin of R2-D2 — is being used to make small deliveries to guests’ rooms. Robots may soon be flipping your burgers or picking the grapes that make your favorite wine.

Video: The Botlr, a robotic bellhop built for Aloft Hotels, delivers an item to a guest in his room.

But the application of robotics is going ever deeper.

The development of robots connected to the Internet, big data, the cloud and advanced computing technology such as artificial intelligence (AI) algorithms are bringing a new class of robots into the workplace — those that can sense, think and act based on specific data and sensory input, and make routine decisions.

In June, the Associated Press began experimenting with having machines write short business stories. The news organization said that eventually, the majority of its U.S. corporate earnings stories would be produced using automation. (As of press time, Risk & Insurance® is not yet employing robotic journalists.)

There are obvious positives to the growth of robotics in the workplace. It makes sense to give robots the high-turnover jobs that are mind-numbingly rote, as well as those jobs and tasks considered extremely dangerous.

But the change that is coming may be far more profound. Garry Mathiason, co-chair of the Robotics, Artificial Intelligence and Automation practice group at Littler Mendelson in San Francisco, cited a 2013 study published by the Oxford Martin School, examining automation potential across the U.S. labor market.

According to the study, said Mathiason, “47 percent of jobs currently done by people in the United States will be done by machines and software within one to two decades. That doesn’t mean there’s going to be 50 percent unemployment; it does mean there’s going to be that much change taking place.”

(For the record, the Oxford Martin study said that insurance underwriters are in the highest risk category for being taken over by automation, just ahead of claims and policy processing personnel, claims adjusters, examiners and investigators.)

“2010 was a turning point in terms of the acceleration of the technology and its implications,” said Mathiason. “There is a change taking place that will be the equivalent of the Internet in terms of what it will do to the workplace.”

Video: At the fulfillment center of North Reading, Mass.-based Kiva Systems, 100 robots work alongside 300 fulfillment associates.

Pointing Fingers

So far though, companies that employ robots see the importance of having human checks and balances on the robots’ work. Many companies are actually increasing staffing levels to support their robotic equipment.

That raises concerns about whether employees are at increased risk of harm by robotic equipment, or may inadvertently interfere with safe robotic operations. A fair number of workplace fatalities related to robotics have occurred in the last decade or two.

As it stands, employers are covered by existing workers’ comp statutes if a robot were to cause a workplace injury or fatality, the same as they would be in the event of an injury or death caused by any other piece of equipment.

Advertisement




The same would not be true, however, if a robot were to injure a customer, a vendor or any other visitor to a facility. In those cases, who can expect to face a lawsuit? The answer, for now, is: It depends.

Liability issues get sketchy when you factor in the element of closed versus open robotics.

In a closed robotics system, a robot is designed and manufactured for one specific purpose. (iRobot’s Roomba, for instance, is a vacuum — period — no matter how many YouTubers use it as an amusement park ride for their cats.)

But with an open robotic system, independent developers would be able to create programs, or apps, to allow the system to accomplish different user-defined tasks, adding more potential culprits in the event of a claim.

“This is going to be a big issue,” said Stephen Wu, an attorney who is of counsel to Silicon Valley Law Group based in San Jose, Calif.

“[In the event of a robot-related accident], I’m going to be doing an investigation and using experts to determine the root cause. Was it my own environment? Was it the hardware manufacturer? Was it the firmware manufacturer? Was it the application software? Was it the operating system manufacturer? Or was it some subcomponents thereof? Or else … was it a data service provider [such as one providing mapping data]?”

These questions will grow still more complicated with the growth of adaptive technology — meaning when robots make decisions on their own based on the data and sensory input they receive.

“Increasingly, adaptive intelligence is being built in where the robot is going to be changing what it does based on external stimuli,” said Drew Haaser, U.S. technology practice leader for Marsh.

“Let’s say it’s putting welds on an auto and it’s been programmed to sense the properties of the materials it’s welding and adapt … . If it makes the wrong decision — what are going to be the legal implications to that?”

That question reaches a whole other level when the consequence is a loss of life.

“What if you have a robot that is … facing the decision to either run over your daughter or hit a school bus full of kids, what is the right thing to do in that situation?” asked David Beyer, managing member of Digital Risk Resources.

“I think that it’s a very complex issue. … They try to think through a lot of these situational risks but you can’t predict all the risks all the time.”

“I think what we’re going to see is that all parties are going to be drawn into these lawsuits. They’re all going to have their feet put to the fire and they’re all going to point at each other. Unfortunately, a jury is going to have to decide these things.” — David Beyer, managing member, Digital Risk Resources

It’s crucial, said Guy Fraker, that we remember robotics is not synonymous with infallible.

“You can say, ‘We can fix that with an algorithm.’ But can you program for that kind of variability in advance? No,” said Fraker, former director of business trends and foresight for State Farm and co-founder and CEO of consultancy Autonomous Stuff.

Advertisement




“We’re learning new things about the technology every day.”

“These are not dumb devices anymore,” said Haaser, “and when they make mistakes in how they interpret stimuli, is it a professional liability error in the design of the product or in the software as opposed to a straightforward bodily injury/property damage claim?”

As more businesses incorporate adaptive technology into the workplace and robotics follow cues based on what they’re learning in their environments, the more that risk management will be expected to have thought through all of the implications of how the robot or robotic systems might respond.

Eventually, there will be cases where robots make “correct” decisions that result in tragic outcomes.

Several experts cited the example of a driverless car faced with a choice of hitting a tractor trailer or hitting an occupied baby stroller. Most assume that the car would attempt to minimize damage by hitting the stroller.

In a case like that, “Was it a mistake? Well, no,” said Haaser.

“Was it what a human with a duty to care would have done? No! And how will the courts treat that? Unfortunately, there are a whole lot of questions and not a whole lot of answers yet as to how the courts will treat that.”

“I think what we’re going to see,” said Beyer, “is that all parties are going to be drawn into these lawsuits.

“They’re all going to have their feet put to the fire and they’re all going to point at each other. Unfortunately, a jury is going to have to decide these things.”

Human Augmentation

The extremely good robotics news is that a great many of the developments coming out of the industry have the potential to decrease employer exposure rather than increase it.

In particular, advances in exoskeleton technology are moving to the forefront, even gaining a global stage during the 2014 World Cup, when a young paraplegic man made the first official kick of the event, wearing a mind-controlled robotic exoskeleton.

Video: Juliano Pinto, 29-year-old paraplegic man, successfully made the first kick of the 2014 World Cup in Sao Paulo, wearing a full body robotic suit.

In South Korea last year, employees of Daewoo Shipbuilding and Marine Engineering helped test a prototype of a full-body exoskeleton that will enable them to lift large hunks of metal, pipes and other objects without excess exertion or risk of strain or injury.

“In basic safety and loss control, the first line of defense is to engineer the risk out; robotics is one of the purest forms of ‘engineering out’ a risk,” said Bill Spiers, vice president and risk consulting manager, Lockton Cos.

Bill Spiers, vice president and risk consulting manager, Lockton Cos.

Bill Spiers, vice president and risk consulting manager, Lockton Cos.

“You have eliminated the human interaction around a task that’s going to create soft tissue strains that are very costly.”

Industry leaders such as Ekso Bionics and Cyberdyne are actively producing exoskeletons that have a broad range of applications. Robotic suits could dramatically reduce injury risk for workers with physically intensive jobs, potentially enhancing productivity at the same time.

“Wearable technology has the potential for ameliorating some [health and safety] concerns,” said Littler Mendelson’s Mathiason. “You can see it someday becoming as common as safety shoes.”

Mathiason said this may eventually be of equal importance as applied to Americans with Disabilities Act accommodation issues.

“You have people who couldn’t perform the essential functions of the job. Then seven million paraplegics can suddenly walk and do things nobody thought they would ever do again,” he said.

For workers who’ve already suffered temporary or permanent disabling injuries, exoskeletons could eventually open the doors for new means to keep injured workers on the job. Even severely disabled employees could potentially be returned to productive and essential work, increasing quality of life for injured workers while saving employers millions in partial and total disability payments.

“You have people who couldn’t perform the essential functions of the job. Then seven million paraplegics can suddenly walk and do things nobody thought they would ever do again.” — Garry Mathiason, co-chair of the Robotics, Artificial Intelligence and Automation practice group, Littler Mendelson

Currently, 330 of Cyberdyne’s HAL-5 full-body exoskeletons have been leased to hospitals across Japan, where they assist patients with muscle weakness or disabilities due to stroke and spinal cord injuries. In some industries, this area of robotics could virtually eliminate obstacles to accommodating injured or disabled workers.

Due Diligence

As companies make critical decisions about incorporating robotics into their operations, it’s important to bring risk management into the loop early on.

While companies increasingly transition processes to robotics for the sake of cost savings, said Haaser, “the challenge for the risk manager will be to see that some of those savings are being reinvested back into risk management.”

Advertisement




But the most urgent piece is getting risk management involved in decisions about capital expenditures like robotics from the get-go.

John Abbott, an account executive at Arthur J. Gallagher & Co., said there are a host of questions that risk managers need to think through in advance.

“Are you going to be using it to do something that’s never been done before? Do you want to go into an area where there’s a potential environmental issue that may impair the robot’s performance? Robots are mechanical and electrical systems — any robotic system is prone to wear and tear and failure to some degree.”

Wu of the Silicon Valley Law Group added that due diligence is of utmost importance when selecting vendors for robotic system components, including whether there are any kinds of certifications that apply to the machines.

“Is there a UL-type seal of approval that we could look for that can be had?” Wu asked. “And what kind of testing went into the robots in the first place?”

Issues such as hackers and the potential for fraud must also be given consideration, said Fraker. “For every great capability that’s ever been developed … there’s an opposing potential dark side,” he cautioned.

10152014_01_cover_story_sidebar

Michelle Kerr is associate editor of Risk & Insurance. She can be reached at [email protected]

More from Risk & Insurance

More from Risk & Insurance

2017 RIMS

Resilience in Face of Cyber

New cyber model platforms will help insurers better manage aggregation risk within their books of business.
By: | April 26, 2017 • 3 min read

As insurers become increasingly concerned about the aggregation of cyber risk exposures in their portfolios, new tools are being developed to help them better assess and manage those exposures.

 One of those tools, a comprehensive cyber risk modeling application for the insurance and reinsurance markets, was announced on April 24 by AIR Worldwide.

Scott Stransky, assistant vice president and principal scientist, AIR Worldwide

Last year at RIMS, AIR announced the release of the industry’s first open source deterministic cyber risk scenario, subsequently releasing a series of scenarios throughout the year, and offering the service to insurers on a consulting basis.

Its latest release, ARC– Analytics of Risk from Cyber — continues that work by offering the modeling platform for license to insurance clients for internal use rather than on a consulting basis. ARC is separate from AIR’s Touchstone platform, allowing for more flexibility in the rapidly changing cyber environment.

ARC allows insurers to get a better picture of their exposures across an entire book of business, with the help of a comprehensive industry exposure database that combines data from multiple public and commercial sources.

The recent attacks on Dyn and Amazon Web Services (AWS) provide perfect examples of how the ARC platform can be used to enhance the industry’s resilience, said Scott Stransky, assistant vice president and principal scientist for AIR Worldwide.

Stransky noted that insurers don’t necessarily have visibility into which of their insureds use Dyn, Amazon Web Services, Rackspace, or other common internet services providers.

In the Dyn and AWS events, there was little insured loss because the downtime fell largely just under policy waiting periods.

But,” said Stransky, “it got our clients thinking, well it happened for a few hours – could it happen for longer? And what does that do to us if it does? … This is really where our model can be very helpful.”

The purpose of having this model is to make the world more resilient … that’s really the goal.”Scott Stransky, assistant vice president and principal scientist, AIR Worldwide

AIR has run the Dyn incident through its model, with the parameters of a single day of downtime impacting the Fortune 1000. Then it did the same with the AWS event.

When we run Fortune 1000 for Dyn for one day, we get a half a billion dollars of loss,” said Stransky. “Taking it one step further – we’ve run the same exercise for AWS for one day, through the Fortune 1000 only, and the losses are about $3 billion.”

So once you expand it out to millions of businesses, the losses would be much higher,” he added.

The ARC platform allows insurers to assess cyber exposures including “silent cyber,” across the spectrum of business, be it D&O, E&O, general liability or property. There are 18 scenarios that can be modeled, with the capability to adjust variables broadly for a better handle on events of varying severity and scope.

Looking ahead, AIR is taking a closer look at what Stransky calls “silent silent cyber,” the complex indirect and difficult to assess or insure potential impacts of any given cyber event.

Stransky cites the 2014 hack of the National Weather Service website as an example. For several days after the hack, no satellite weather imagery was available to be fed into weather models.

Imagine there was a hurricane happening during the time there was no weather service imagery,” he said. “[So] the models wouldn’t have been as accurate; people wouldn’t have had as much advance warning; they wouldn’t have evacuated as quickly or boarded up their homes.”

It’s possible that the losses would be significantly higher in such a scenario, but there would be no way to quantify how much of it could be attributed to the cyber attack and how much was strictly the result of the hurricane itself.

It’s very, very indirect,” said Stransky, citing the recent hack of the Dallas tornado sirens as another example. Not only did the situation jam up the 911 system, potentially exacerbating any number of crisis events, but such a false alarm could lead to increased losses in the future.

The next time if there’s a real tornado, people make think, ‘Oh, its just some hack,’ ” he said. “So if there’s a real tornado, who knows what’s going to happen.”

Modeling for “silent silent cyber” remains elusive. But platforms like ARC are a step in the right direction for ensuring the continued health and strength of the insurance industry in the face of the ever-changing specter of cyber exposure.

Because we have this model, insurers are now able to manage the risks better, to be more resilient against cyber attacks, to really understand their portfolios,” said Stransky. “So when it does happen, they’ll be able to respond, they’ll be able to pay out the claims properly, they’ll be prepared.

The purpose of having this model is to make the world more resilient … that’s really the goal.”

Additional stories from RIMS 2017:

Blockchain Pros and Cons

If barriers to implementation are brought down, blockchain offers potential for financial institutions.

Embrace the Internet of Things

Risk managers can use IoT for data analytics and other risk mitigation needs, but connected devices also offer a multitude of exposures.

Feeling Unprepared to Deal With Risks

Damage to brand and reputation ranked as the top risk concern of risk managers throughout the world.

Reviewing Medical Marijuana Claims

Liberty Mutual appears to be the first carrier to create a workflow process for evaluating medical marijuana expense reimbursement requests.

Cyber Threat Will Get More Difficult

Companies should focus on response, resiliency and recovery when it comes to cyber risks.

RIMS Conference Held in Birthplace of Insurance in US

Carriers continue their vital role of helping insureds mitigate risks and promote safety.

Michelle Kerr is associate editor of Risk & Insurance. She can be reached at [email protected]