Part 1 of an occasional series on automation, robots and AI.
In the robotics lab at the University of Louisville, a powerful-looking robot called Baxter recently struggled with a simple task: transferring an empty plastic cup from one claw to the other.
The pincers attached to the end of Baxter’s bulky, red, left arm grasped the cup from a table without a hitch, but as the pincers on its right arm approached, they bumped into the cup and moved it into a near-horizontal position, requiring the intervention of a human graduate student, who placed the cup back on the table.
Somewhat pitifully, Baxter, in the firm belief that it was still holding the cup, moved its right arm carefully back to the table to complete the motion, empty as its pincers were. The robot hadn’t even noticed that the as-yet superior human already had completed the task.
Baxter fails the cup-switching task about 20 percent of the time — far too high a rate to be able to employ him as a barista. Never mind that the robot moves with such apathy that any customer in desperate need of a jolt of java might hurry out of the coffee shop in frustration.
While a cafe in Japan this year began dispensing coffee with the help of a robot, Baxter’s struggles point to the long way that robots still have to go before they can perform anything but the most rudimentary human tasks, said Dan Popa, the Vogt Endowed Chair on Advanced Manufacturing in the university’s Department of Electrical and Computer Engineering.
“It’s true that robots will take over,” Popa said, “but they will take over the mundane and boring jobs. Not everything is going to get automated.”
However, Popa’s UofL colleague Roman Yampolskiy, an artificial intelligence expert, disagrees.
The robots aren’t coming, he said, they’re already here. And it’s only a matter of time before they’re coming for your job.
“It’s only getting worse,” said Yampolskiy, associate professor at the Speed School of Engineering and director of the university’s Cybersecurity Laboratory.
Eventually, robots or AI systems, or both, will perform every task better than a human, he said, and only the complexity of your work may give you temporary reprieve.
Popa and Yampolskiy represent the two hemispheres of the brain (or, if you will, halves of the positronic cortex) that describe the future of work and humanity alongside (or under?) increasingly more powerful and capable machines:
In one hemisphere is the R2D2 camp. As in tech revolutions of the past, new technologies mean new jobs will replace the old ones, much like jobs in the horse-drawn carriage industry were (more than) replaced by jobs in the automobile industry. In addition, technological advances create jobs that did not exist before, for example, a century ago, few people worked in search engine optimization.
In the other is the T-1000 camp, which believes that the previous cycle will not continue indefinitely, and, that someday advanced machines and AI will take over all human jobs.
For sure, robots and intelligent systems are encroaching ever more into human workspaces, in sectors as varied as health care and manufacturing. While at this point, the machines are often enhancing the work of their human colleagues — they sometimes also are replacing them altogether:
- Areas of Ford Motor Co.’s Kentucky Truck Plant require nary a human, as robots twist and turn to lift and carry truck hoods with ease. “Kong,” an aptly named monster of a robot, lifts entire truck frames that weigh thousands of pounds.
- At Norton Healthcare, doctors perform more precise brain surgeries thanks to the mechanically steady “hands” of a robot. Some of the procedures wouldn’t be possible — or advisable — without the machine.
- Software and scanners at UPS allow for packages to be moved and rerouted over miles of conveyor belts without the touch of a human hand. Similar high-tech machinery allows Humana to more efficiently mail hundreds of thousands of prescription drugs to customers’ homes.
- Customer service reps at Humana are getting instant feedback about callers’ emotional states from an artificial intelligence assistant.
- And at UofL, Popa is developing, with the help of a federal grant, a robot that will interact with and monitor hospital patients, not to replace nurses, but to allow them to focus on more complex tasks.
Robots do some things very well, Popa said, because, like Ford’s “Kong,” they possess greater strength than humans, and because their steady, mechanical “hands” allow them to perform tasks such as placing a small neurological laser into a certain part of the brain, with greater precision — but before they become ubiquitous, they have to clear several hurdles, including safety, capability, consistency and cost.
Even the impressive and intimidating robots from Boston Dynamics, which show four-legged machines opening doors, and a two-legged one doing a somersault, still are performing only “low-level tasks,” Popa said.
They still struggle with other relatively simple, but higher-level actions that require intelligence and perception, such as adjusting to the unexpected or figuring out whether they can successfully negotiate a certain terrain, he said.
Those are critical skills that keep the machines from performing anything but routine tasks in controlled environments, Popa said. The developers can’t risk losing million-dollar robot in a swamp, for example.
“There are many things that robots can’t do,” Popa said.
And even for the things that they can do, their high rate of failure (see Baxter’s barista skills) and breakdowns limit their uses: Performing a task satisfactorily 80 percent of the time is not going to suffice on a the refrigerator production line at GE Appliances, for example.
“Also, if you need a human to chaperone the robot, there’s no point in having (it),” Popa said.
Robots’ usefulness also is being curbed by the machines’ inability to learn: Any time a mechanical process changes, the robot has to be reprogrammed by a human.
“Robots can’t really do anything on their own,” Popa said.
The machines’ high cost also is hindering the rise of the robots. While the price of sensors and processors has fallen, the cost of other physical components — the mechanics, the actuators — has remained steady, which makes use of robots cost-prohibitive in many applications.
Popa said concerns about safety, too, keep capable robots out of factories and homes.
While humans have allowed small robots with limited strength and uses into their homes, to do vacuuming, for example, they are likely to have considerable reservations if a humanlike robot with Hulk-like strength strides into their kitchen and grabs a 12-inch Gyuto Japanese chef knife — even if it’s just to carve the Thanksgiving turkey.
Security and safety concerns rise with a robot’s proficiency and power, Popa said. And any time you give a robot a potentially dangerous item, such a knife, an ice pick or a light saber, the robot will be less safe — and its human owner significantly less inclined to be anywhere near the machine.
Popa said people often also forget about how the use of robots can increase human safety — and job retention.
Without productivity gains from mechanization, production of more goods likely would have been moved from the U.S. to low labor cost countries in the last few decades, he said.
“Robots actually help keep production here,” Popa said.
In other cases, robots are performing dangerous work and therefore protecting the health and safety of humans by tackling tasks that are highly repetitive or performed in potentially dangerous locations, he said.
And while automation is replacing some jobs, robots and intelligent systems also are creating jobs for designers, programmers, engineers and service technicians.
Robots and intelligent systems eventually may occupy all jobs, Popa acknowledged.
“But not anytime soon,” he said. “Not in our lifetime.”
*Past performance is not an indicator of future results
For Yampolskiy, the AI expert, it’s clear that robots and artificial intelligences eventually will take over all human jobs, and one of the few questions that remain is what humans should do about it.
In the last few decades, many American jobs have been displaced internationally, with companies moving production to low-wage countries with few environmental and worker protection laws, he said.
With continuing technological advances, American jobs increasingly are being replaced by robots and software: From bank ATMs and automated grocery checkout lanes, to cashier-less stores, to virtual phone assistants, manufacturing machines and autonomous vehicles, humans’ work — and failures — increasingly are giving way to superior AI and machinery.
“That’s only going to get exponentially worse,” Yampolskiy said.
Jason Latta, a Louisville entrepreneur and futurist, agrees.
While automation has threatened primarily blue-collar jobs, AI systems also threaten white-collar occupations, including paralegals, accountants and insurance claims adjusters, the tasks of whom include researching legal cases, combing through financial data and analyzing insurance claims against rule sets.
“Those are all automatable tasks,” Latta said. “AI is definitely going to have an impact on white-collar jobs, even more so than people expect.”
While predictions about how much robots will affect human jobs vary widely, agencies including the Organisation for Economic Cooperation and Development and the World Economic Forum said that tens of millions of jobs will be lost in the U.S. alone within the next decade or so.
McKinsey Global Institute said in December that it expects about a quarter of current work activity hours in the U.S. to be replaced by 2030 and about a third of the workforce “may need to switch occupational groups.”
Worldwide, McKinsey said, “between 400 million and 800 million individuals could be displaced by automation and need to find new jobs.”
The losses will hit especially the people with predictable physical work, such as record clerks, equipment installers and food preparation workers, McKinsey said.
The analyses by McKinsey and others indicate that automation also is going to create jobs — though fewer than it is eliminating.
“Workers of the future will spend more time on activities that machines are less capable of, such as managing people, applying expertise, and communicating with others,” McKinsey wrote. “They will spend less time on predictable physical activities and on collecting and processing data, where machines already exceed human performance.
“The skills and capabilities required will also shift, requiring more social and emotional skills and more advanced cognitive capabilities, such as logical reasoning and creativity.”
And that’s a problem, Yampolskiy said, because lower skill jobs are disappearing and being replaced with higher skill jobs, thereby exacerbating an already existing problem: the gap between the skills that many of today’s jobs require and the skills that many of today’s applicants possess.
“We are creating new jobs,” Yampolskiy said, “but they are much more technically demanding.”
All the 50+-year-old truck drivers whose jobs will be replaced by self-driving trucks are unlikely to become AI developers, he said.
Popa said that at least in the near and intermediate future, robots will continue to make humans’ lives and jobs easier.
For example, Popa and his team at UofL soon plan to release their nursing assistant robot, which will serve as a patient sitter, monitor and walker.
The robot will be able to bring patients items such as a cup of water or a remote control, monitor movements and vital signs and alert a (human) nurse or doctor if a patient suffers a medical problem. The robot also will accompany patients on walks down hospital corridors while carrying IV bags and other items.
Those tasks are important, because they aid in patient outcome and recovery, Popa said, but they currently have to be performed by (expensive) human patient sitters. The robot also would allow nurses to dedicate their time to more important tasks, which may help mitigate the nursing shortage, which is expected to get worse as more people reach old age and require care. The Bureau of Labor Statistics estimates that by 2024, including expected retirements, the number of new nurses who will be needed in the U.S. exceeds one million.
Popa’s Adaptive Robotic Nursing Assistant is being funded in part by a $1 million grant from the National Science Foundation. UofL is collaborating with a startup in Pittsburgh. The ARNA will have one arm, to reduce costs, Popa said. The prototype’s components alone cost $60,000, but if thousands of hospitals express interest, costs would decline with ramped-up production.
“And eventually you may want to gift a robot like this to your parents,” Popa said.
Robots already are helping humans in the medical industry. For example, Popa has conducted research that showed robots can detect autism earlier than humans and also can teach social skills to autistic kids much better than humans can, because the kids relate to technology a lot better than to other people.
Yampolskiy acknowledged that robot-induced job losses, even on a global scale, may not necessarily be a bad thing, so long as societies can respond effectively to potentially massive unemployment.
“Most people hate their job,” he said.
If robots replace people in boring, crappy, low-wage jobs, “it might be a great improvement.” Freed from the shackles of their stupefying jobs, people may find new ways to be creative and productive, Yampolskiy said.
“There are some things you can look forward to,” he said.
Yampolskiy also advised that people who are in school or early in their careers should choose relatively robot-proof professions so that they may escape, at least for a while, the occupational effects of the rise of the robots.
“Nobody thinks their job is going to be automated,” Yampolskiy said, but people should ask themselves how repetitive their current or future job is and how long it might take to train someone else to do it.
If your job is highly repetitive, much of your work is spent in one location and it takes only a couple of days to train someone to do it, Yampolskiy said a robot probably already has its scanners on you.
Robots eventually will replace every worker, Yampolskiy said, but young people can make better career decisions to increase their chances of remaining employed for the foreseeable future.
“You don’t have to be an early victim,” he said
Join Insider Louisville July 24 for Rise of the Robots, a discussion with local experts, including Popa and Yampolskiy, on the ways in which robots and AI are affecting the local economy.