Listen to this article
“Cobots”, or collaborative robots, are making inroads into work previously considered too difficult to automate. But as cobots get better at performing tasks such as material handling or packaging, their designers are having to consider the effects on their colleagues of the machines’ improved ability to interact with humans.
In its early stages, this new technology has been safe if underwhelming, says David Mindell, a professor at Massachusetts Institute of Technology. Of the cobots, he says: “They don’t do much collaboration, but at least they won’t cut your head off.”
Small, light and slow moving, cobots are generally harmless — the sensors and machine-learning software that enable them to “understand” their environment have a simple override: if a human gets too close, they are programmed to shut down.
The first job has been to design the software models to allow robots to operate in the human world, says Manuela Veloso, head of machine learning at Carnegie Mellon’s School of Computer Science. “It’s very important to be able to envision a mobile creature moving around in our space,” she says. For instance, getting machines to work alongside people will require an understanding of “safety zones” of the body: “We’re trying to model a person. You don’t want to hit an eye — an elbow is less important.”
As the software becomes more sophisticated, it promises more flexible machines that can be released from their cages. “We’ve got people doing jobs today because the regular robots can’t do it,” says Jim Lawton, head of product and marketing at Rethink Robotics, a Boston-based maker of cobots. These often involve repetitive actions that strain human limbs, are mind-numbingly dull and consign workers to jobs with no chance of career advancement, he says.
Mindell, author of Our Robots, Ourselves, a 2015 book about human-robot interaction, agrees there is much to be gained in the way of worker wellbeing: “If your work is truly about to be augmented, or made less dangerous or less straining, these are good things.” But he says that limits in both the technology and imagination on how to apply it have made this more promise than reality.
Designing complex interactions between robots and people will take a change in mindset, he says, adding that the history of automation has largely been about treating humans like robots, to fit into automated processes. “The computer science world still has a long way to go before it has a clue about how to deal with people,” he says.
At a simple level, makers of cobots are working to reduce the sense of weirdness for people working alongside machines whose level of intelligence they find hard to judge. Rethink, for instance, experimented with putting smiling mouths on its robots to make them seem more “human”. The result was the opposite, says Lawton: people thought the machines were smirking at them, and found them “arrogant and condescending”. Moving into the “uncanny valley” where robots start to copy humans too closely “spooked people”, he says.
Veloso says there are hurdles that will have to be overcome to improve the human experience of working with the machines. One is that the machines have to be more understandable. “The more humans infer what a robot will do next, the safer it will be,” she says.
Rethink’s answer has been to give its robots “eyes” (an image on a tablet computer) that indicate the direction the machine is about to move in — a simple way to prepare people around them that they are about to do something, says Lawton.
Another key is to design a form of robot-human symbiosis in which each helps the other achieve its goal, says Veloso. That will mean teaching people to respond to requests from the robots, or to anticipate their needs, as much as the other way around. As interactions like this become more subtle and machines take over more work alongside people, the long-term impact on the wellbeing of human workers is hard to predict. Against the obvious benefits of taking dangerous or tedious work away from people, there may be unexpected side-effects. “When people invented keyboards, they weren’t imagining carpal tunnel syndrome,” Veloso points out.
As more automation creeps in, there may be subtle but far-reaching effects on the way work is designed. There is a fear that the iterative process improvements that are a product of lean manufacturing — constantly learning and implementing better ways of working — may be threatened, says Lawton. If existing work processes are automated, the result could be an ossific9ation that prevents this steady improvement.
Like much technology whose benefits are clear in the short term, even if their long-term effects on human wellbeing are hard to judge, the advance of the cobots is unlikely to be slowed. People are likely to take to their new robot colleagues as enthusiastically as they took to their smartphones, says Mindell. “People have their fears — in some ways, they are legitimate fears,” he says. “At the same time, they are addicted to their technology.”
‘Algorithms took our jobs’
Tom Gordon was 45 when his lucrative career as an oil trader suddenly faced a new threat. Electronic trading, which originally had been introduced to expand trading capacity overnight, was now operating head-to-head with Gordon and his colleagues on the floor of the exchange during the day.
Gordon says he used to handle between 500 and 750 trades a day. In his nearly 25 years as a trader he recalls recording only two months of losses. But even the high volumes that a successful trader like Gordon could handle were quickly overshadowed by the volumes electronic systems were capable of processing.
For Gordon, working alongside the electronic market was like being hit by a truck. “I saw the transition was coming and knew [traders] were going to get run over,” he says. He eventually left and retrained as a social worker.
He was wise to do so, because a few years later, in 2016, CME Group, which owns the New York Mercantile Exchange (Nymex), closed the last of its remaining commodity-trading pits.
Gordon says some of his former colleagues have struggled to cope in their new lives. “Some have done quite well, but for many of the people it really broke their lives and their spirit.”
Losing a job to a machine or algorithm carries a unique psychological burden, says Marty Nemko, a psychologist and career counsellor.
No training exists that can help a human match the speed and efficiency of artificial intelligence. “There is an inevitability of [one’s] inferior ability that accrues,” Nemko says.
Tim Leberecht, a consultant on business leadership, agrees: “If we lose our jobs due to automation and can’t get back into the workforce, then there is this huge void of purpose and meaning.”
“The big issue with this fourth industrial revolution is that we don’t have the social institutions that are facilitating and enabling the transition,” says Ravin Jesuthasan, managing director at Willis Towers Watson, and leader of the consulting group’s research area, “Future of Work”.
Research on the threat of automation paints a complicated picture. A 2016 OECD report found an average of 9 per cent of all jobs across the 21 countries the research covered could be automated, given current technology. A report by consultants McKinsey puts the global figure at less than 5 per cent.
Many researchers suggest the more nuanced effect of this transition will be on the handful of tasks across all sectors that are routine and repetitive.
According to another McKinsey report, more than 70 per cent of tasks performed by workers in the food service and hospitality sector could be carried out by machines. In manufacturing, nearly 60 per cent of tasks in jobs such as welding and maintaining equipment are at risk.
Higher-paying jobs are not immune from the disruption. McKinsey found that up to 50 per cent of tasks in the financial services industry could be automated, as could about a third of jobs in healthcare.
Jesuthasan says this refocusing of tasks can give people the space to do more meaningful work. “Leaving behind all of those routine things [creates] a huge emphasis on creativity and empathy and care,” he says.
After witnessing his original job as a trader vanish, it is perhaps no surprise that Gordon has found himself engrossed in work requiring these human characteristics. “I want to do my part,” he says. “Will I make a difference? I don’t know, but I’m going to give it a shot.”
Get alerts on AI and jobs when a new story is published