Health

Can Robots Ease the Caregiver Burden?

A small robot army is being test-driven in long-term care facilities throughout Toronto to fill an urgent need—to look after our elderly. But what is lost when we outsource caregiving to machines?

illustration of round-headed white robots holding up cups of tea and sandwiches

Illustration, Ryan Snook.

Brian sits at a table across from an elderly woman. He is wearing a blue T-shirt and a baseball hat. She is petite, with curly grey hair, an expectant smile, and a tray of scrambled eggs in front of her. “The side dish looks very tasty,” he says in a cheerful primary school teacher voice. “Why don’t you try some?” She takes a bite. “It’s pretty good, Brian,” she says politely.

Brian’s own smile is a little robotic, which isn’t judgey; he’s a prototype socially assistive robot. His stiff expressions are created via actuators beneath his silicon rubber face, and his blue T-shirt is draped over a torso of wires and arms with metallic hands. The full effect borders on creepy, a mash-up of an emoji, an over-plugged outlet and a mannequin. Yet, among the seniors who encounter him in research testing, Brian is a hit.

Brian’s job is to motivate older adults to eat. Compared to robots of the collective imagination, a food-coaxing droid seems a little mundane; Brian is but a distant, unacknowledged cousin of the Terminator. But eating can be a life-and-death health issue for elderly people in care, especially those with cognitive impairment, and that population is the focus of the work under way at the University of Toronto’s Autonomous Systems and Biomechatronics Laboratory (ASB Lab), where Brian and his fellow prototypes have been developed. Goldie Nejat, Canada Research Chair in Robots for Society at U of T, and her research team are building an array of socially assistive robots that can offer support for everyday and repetitive tasks such as eating, selecting clothes or exercising, with the aim of lightening the burden for seniors and their caregivers. And the future is closing in: Nejat predicts that robots could become a common presence in care facilities within two to five years.

The technology that animates Brian and his ilk might be complex and anchored in artificial intelligence, but the promise of “robocarers”—a field that’s booming around the world—is simple. Robots give us time. If these machines take on much of what sociologists call the “dull and dirty” work of caregiving, humans will be freed to do what matters most: companionship, touching, talking. The caring part of caregiving will be restored.

After Brian has made a couple of bad jokes (“Why did the cookie go to the doctor?” “She was feeling crummy!”), and the woman has eaten her eggs and finished her juice, their time together is done. Sweetly, she says: “Thank you, Brian. I hope to see you again.”


To observe a person extend such graciousness to a machine is strange, and somehow poignant, yet most of the subjects who meet Brian do so. Watching the video of these trials, it’s hard not to interpret the seniors’ social largesse as a yearning for company and attention. Perhaps that’s because the inevitable integration of robots and caregiving doesn’t just signal a triumph of technological innovation but also serves as a stark reminder of our failings when it comes to the elderly. It’s clear that the robots are on their way, but the desired outcome is much murkier. Who wins, and who loses, when the machines arrive to do what we can’t?

When my grandmother was nearing the end of her life, and moving through the stages of dementia, she lived in a government-funded complex-care home in Vancouver. It was a pretty, wood-sided house with a tidy green lawn and warm staff—about as good as long-term care can get. Yet, I remember sitting with her in her room when she was agitated, hitting the call button and waiting with frustration for someone to arrive. Perhaps whoever was on call was tired; surely, they were all overworked. Still, waiting felt like an affront. In the eyes of a loving family member, the attention provided could never be sufficient. No caregiver could give her enough support, enough love, enough time.

The experience of bearing witness to age and illness (let alone going through it) is a profoundly personal experience. But we tend to think of the upcoming surge in the elderly population as a monolith, a faceless demographic shift. In Canada, the aging baby boomers are skewing our collective age upward. In 2014, over six million Canadians were seniors, aged 65 or older, representing 15.6 percent of the population. By 2030, seniors are projected to number 9.5 million, making up 23 percent of Canadians. The old will be older, too, with life expectancy increasing from 84.2 to 86.2 for women, and from 80 to 82.9 for men by 2036.

Resources to serve them are scarce and patchily available. Medical advances mean many will be living with chronic illnesses. They’ll likely want to age at home for as long as possible and public policy shifts encourage it, even with shortages in personal care workers. Retirement communities, which include long-term care facilities, are costly—monthly fees range from $1,400 to over $4,500 across Canada—and wait lists and bed shortages are the norm. In Ontario, the average time from request to placement in long-term care is 161 days.

Whether by choice or circumstance, friends and family members will often step in to meet the unmet needs of loved ones, but there are limits to this semi-solution. Thirty-five percent of working Canadians balance jobs and caregiving responsibilities to the detriment of their mental and physical well-being, as well as the economy’s: Unpaid caregiving is estimated to cost employers $1.3 billion annually in lost productivity, absenteeism and turnover.

The tech sector is poised to help—and profit from—the age of aging. A Belgium-based company called ZoraBots has sold “Zora” miniature companions to nursing homes in Europe, the U.S. and the Middle East. China is sinking millions into robocarers to address the conundrum of a massive aging population and a one-child policy that’s thinned the ranks of family caregivers. Throughout Japan, it’s not unusual for senior facilities to use automatic feeding robots, assistive devices that chat with Alzheimer’s patients, or the “Robear”—a polar bear-like creature that lifts and moves frail patients between beds.

According to the International Federation of Robotics, sales of personal service robots increased 30 percent between 2016 and 2018, to reach almost US$1.6 billion. Right now, they’re prohibitively expensive: The knee-high, dancing and game-playing Zora robot, rolled out in some nursing homes in France in 2018, can cost up to US$18,000. But as demand and production increase, the price is expected to drop.

In Canadian health care, humanoid robots are still relatively rare, though cheaper robotic pets are used as ostensible therapy animals in some homes and clinical settings. The most well-known is Paro the seal, an artificially intelligent fluffy automaton developed in Japan that costs about $6,500 and can bleat and flap its flippers in response to a user’s touch, offering company and empathy to Alzheimer’s and dementia patients.

In the midst of this tech-care explosion, a small robot army is being test-driven in trials at the University of Toronto and at retirement and long-term care homes throughout the city. Besides Brian, an early prototype, the robot fleet includes Tangy, who can host bingo games, and Casper, who can guide a senior through the steps to make a basic meal. As the technology evolves, a single automaton will be able to do all these tasks, but for now, they’re still designed as one-trick robots.

Perhaps this seems underwhelming. Stoked by sci-fi, the public imagination might expect that robots are advanced enough to hang out with Nana tomorrow, but building a machine that’s emotionally sensitive is still a major research problem. Paro has fairly primitive emotional intelligence; its responses when touched seem hardly more advanced than a Teddy Ruxpin. The tasks that Japanese robots are performing in nursing homes today are largely physical. Canadian researchers, including Nejat, are distinguishing themselves in the robotics field by focusing on robots that can assist a vulnerable population by learning and adapting from the user in the way that human caregivers do with older adults. Nejat has been doing this for 15 years, and the field is just starting to show significant advances.

In a vast, mostly empty room overlooking the city at U of T’s downtown campus, Nejat and two team members introduce me to Salt, a gleaming white robot with an egg-like head. Less than four feet tall, it resembles a precocious toddler—a design decision to reduce intimidation.

Salt’s body, the hardware, comes from the Japanese company SoftBank (one of the lab’s many partners) but the intelligence is developed at U of T so it can be transferred between platforms, perhaps to be used inside a different robot. This particular Salt is designed to lead seniors in upper body exercises.

“Robots can repeat the same mundane task over and over again with a smile on their face and a happy emotion. The robot never gets tired of doing it,” says Nejat.

Salt is placed at a very specific distance from me—between 1.5 and 2 metres—because researchers have found that any closer feels like a violation of social space. It moves its head encouragingly, and lightly sways its arms and waist; both movements subconsciously register as human. Salt is designed to be gender neutral, though a female voice instantly makes me view her as female. Still, Salt isn’t exactly passing as a person, and that’s strategic, too: A robot that’s too human is off-putting. (That dissonance between a near-perfect simulation and an actual person can cause an eerie, even repulsed feeling that’s been called “the uncanny valley,” and no developer wants to end up there.)
In a voice laden with exclamation points, Salt says: “We are going to do our exercises together! Are you ready?” Salt takes in the user’s speech, facial expressions, gestures and body language, and adapts fast. The bot is sensitive enough to respond to the mood of the user by changing eye colour: a positive valence, and the eyes are green; if the user is talking, they turn yellow to show she’s listening and interested.

“Let’s get started,” Salt exclaims. I immediately answer: “Okay, yes!” It’s hard not to want to please Salt; she’s very peppy. Nejat’s researchers have found that humans are, overall, remarkably accepting of social robots. In user trials, compliance and acceptance levels are consistently high; older adults and family members flock to the robo curiosities.

Nejat and her team repeatedly do field research in long-term care facilities to figure out what engineers should be prioritizing. They observed that language barriers in multicultural Toronto were an issue; hence, the robots they’re developing can speak different languages (and they’re currently working on multilingual speech recognition). Some caregivers with packed schedules said they wanted the robots to offer them task prompts, like when to move a patient, so reminders are built into the behaviour. Others said that it would be helpful if a robot would simply sit next to a patient to give them someone to talk to. Multiple user studies follow every technological breakthrough. The caregivers, Nejat has noted, are dedicated, competent—and overworked. “We would sometimes see the struggles of caregivers to get people to eat, to get dressed. Meanwhile, the residents were getting no stimulation.”

Growing up in the suburbs of Toronto, Nejat was the kind of mechanically minded child who used software to build robots for fun. When she entered the robotics field in the early 2000s, the baby boomers were beginning to retire and dementia was on the rise. Robots, she believed, could help address the scarcity of people with the skills to look after the oncoming swell of seniors, improving quality of life for both caregivers and patients. “It doesn’t have to be robots that do the task for you, but robots that support you to do the task,” she says.

As I followed Salt’s movements, doing my arm bends and receiving my praise (“It was fun doing this exercise with you!”), I admit that part of me wanted to knock her over just to see how she would react. This response is rare, researchers assured me, somewhat alarmed, and Salt will stop and ask a user if everything is okay if sensors register that someone isn’t participating. I could imagine Salt-ercise as a useful activity in a busy nursing home. But I also thought that a human occupational therapist would be able to touch my arms, adjust my posture, maybe hug me. I remembered how much my grandmother responded to touch; how my mother would hold her hand, or rub lotion onto her arms, or recite a poem that my grandmother remembered. These moments were profoundly human, personal. Mediating them with a robot seems unimaginable.

Nejat understands the hesitancy. “If we had enough caregivers, robocare wouldn’t be happening. But unfortunately we don’t—so what do we do? Do we let people sit there and watch the elevator doors open all day long? Or can we help and support them with technology?” she asks. “If we had the human power, then we would all be working on building robots for manufacturing or other services. But you’re seeing people—especially those with dementia— sitting there day by day, and symptoms getting worse if they don’t have the stimulation.”

It’s true: The robots are rising to fill a hole. Our universal healthcare model, if imperfect in execution, generally assures us that the urgent medical conditions of the elderly will be treated. But what to do with seniors in the time between appointments at the hospital or doctor’s office is less straightforward, and it’s largely left to families to take on that labour themselves or seek help on their own. When an aging loved one requires support—or we can no longer function independently ourselves—there is no road map, and the stress of careening into this stage of life can be overwhelm- ing. Paid care workers experience high levels of burnout, and there are staffing shortages. In Canada, unpaid, informal caregivers (i.e., relatives) spend, on average, 19 hours a week on caregiving duties—with one in 10 devoting more than 30 hours a week— resulting in profound psychological and physical distress.

I look at Salt, standing at attention, and I think of my grandmother, and Nejat’s question: “What else are we going to do?” Well, we could train and hire more personal support workers, and then proclaim the value of this work (mostly performed by women of colour) by paying better wages. We could strengthen employment laws to protect family members (again: mostly women) who take time off work to care for their loved ones. We could follow the recommendations of the Canadian Medical Association to implement a national seniors strategy, making critical investments in long-term and nursing care to avoid a looming crisis of wait times and bed shortages.

Or, you know . . . robots.

Donna Thomson, the Ottawa-based co-author of The Unexpected Journey of Caring, has a lifetime of experience caregiving. Her father had three strokes when she was a teenager and her son, Nicholas, has cerebral palsy and severe physical disabilities. Thomson’s mother died in 2018 at age 96. Though Nicholas entered a care home in 2011, Thomson and her husband provided him 24-hour nursing support with very little paid help for 23 years. When robot carers started getting media attention, she was curious and conflicted. A phrase that Thomson had heard in the community defines caregiving as “love in action.” Her first instinct was: “A robot cannot love.”

And yet Thomson knows first-hand that many caregiving activities are “dull and dirty.” “Not every action needs to be rooted in love. Bring me my pills, change the sheets, cut the grass, make the dinner—they are just tasks to be performed. There might be a role in those tasks for technology,” she says. “The danger is when technology is a substitute for intimacy. A dear friend’s mother-in-law died last week, and they were all in bed with her when she died— five people in bed holding her. No robot is going to do that.”

We imagine that off-loading rote tasks means creating a space in the day where we can be fully present with our loved ones. But what would we do with that time, exactly? The tasks that my mother did with my grandmother when she visited daily—selecting her clothes, tidying her room—didn’t seem to interfere with their relationship; they became the relationship. The disease had put limits on their interactions. Doing commonplace things became ritualistic, a way to be together.

Humans are social animals, hard-wired for caring, which releases endorphins, creating what psychologists call a “helper’s high.” Jean Vanier, the late Catholic theologian and disabled rights pioneer, argued that caregiving flows two ways, deepening each party’s humanity: “The weak teach the strong to accept and integrate the weakness and brokenness of their own lives.” If fundamental, routine tasks are farmed out to technology, perhaps we miss an entry point to the best of ourselves. Perhaps we, as humans, are changed.

Still, for those caught in the caregiving squeeze, such philosophical questions are a luxury; any respite would be welcome, in any form. Robots can, theoretically, provide caregivers that relief.

Less known is what effects robocare will have on patients. Research on long-term relationships with robots is scant. Australian and German researchers found that patients with dementia who spent time with Paro the therapy seal reported improvements to quality of life, and their anxiety levels dropped substantially. In other words, the benefits of an electronic seal that responds to touch and voice aren’t dissimilar from the benefits of social contact at any age. Sociability is, in fact, a key to aging well and the best insurance against dementia. Loneliness and social isolation among seniors are linked to increased health risks like obesity and even early death.

With that in mind, nursing homes are primarily using robots for companionship. One U of T study showed that visits with Tangy, the bingo-hosting robot, in a long-term care facility improved sociability among a group of elderly people in the home—and not solely because they were interacting with Tangy, but because they began to speak with one another about it.

But robots’ sociability could also be a liability. Some medical ethicists are concerned that robots may in fact reduce the amount of human contact a patient receives. If a robot leads an exercise class, or changes the bedpans, some patients without family or friends might lose crucial points of human contact.

“The insult to our dignity isn’t getting the seal,” says Thomson, referring to Paro. “It’s when we say: You get the seal and no visitors.” As we age, we often lose control over our lives, and our autonomy becomes compromised by immobility. Tom Kitwood, the late British pioneer of dementia research, wrote about the risk of objectification when a human is cared for by a robot: “Objectification—treating a person as if they were a lump of dead matter: to be pushed, lifted, pumped or drained, without proper reference to the fact that they are sentient beings.” In this dark, Black Mirror scenario, elderly people are reduced to things handled by things.

Regardless of such trepidation, technology is already reinventing healthcare. Eric Topol, an American cardiologist and geneticist, believes that patients of all ages will be the better for this transformational shift—if it’s done the right way. In his new book, Deep Medicine, about AI and healthcare, Topol argues that if algorithms can supplement much of the labour that doctors expend, particularly on diagnosing images, doctors can return their time and attention to the patient. Topol told me that those looking after elderly family members may be similarly liberated by an algorithm or machine that can keep track of medical data better than any human. But work has an amazing ability to take up all the space—it is the goldfish that grows to fit the bowl—and if caregivers or doctors suddenly have more time, it seems entirely probable they’ll just fill it with even more tasks, by instinct or institutional obligation.

“If we don’t use this new efficiency provided by technology to the advantage of patients, then this remarkable, unprecedented opportunity will be for naught,” says Topol. “As bad as it is with burnout, it could be worse if we don’t use this new-found time in the right way, by turning inward and giving back time to the patient-doctor relationship, or the patient-caregiver relationship, which have been eroding for decades.”

Fifty percent of the global workforce could be automated by 2055, according to the McKinsey Global Institute. The overwhelming majority of caregivers for the elderly—up to 81 percent—are female, meaning that jobs for women are particularly vulnerable to the robot revolution. Nejat is adamant that the robots will not displace, but will work in concert with, medical professionals and caregivers.

But as robocare becomes cheaper, and more prevalent, a dark future scenario emerges in which human caregiving might become a luxury item. A live companion or nurse—the gold standard for care—will be reserved for the wealthy, while the rest of our seniors are handed electronic seals.

Earbuds, smartphones, Alexa, the Roomba—we have so far shown a shrugging willingness to allow the mechanical to intervene in our experience. But if there is a line, a point where we say that technology isn’t solely a boon, it might be at caregiving, the most intimate, important work of being human.

Allison Mathieson is a former preschool teacher in Toronto who agreed to participate in one of the U of T robot studies. She spent an afternoon with Casper, a squat fellow with a tablet on his chest who led her through the steps to make a tuna sandwich.

Mathieson felt a little nervous the day she went in to the lab, but fascinated, too. With the robot’s prompting, she got the necessary items out of the fridge, opened the tuna tin, mixed in the mayonnaise—the kind of sequencing that can be hard for someone with cognitive impairment. Casper was a good guide, she said, even kind of fun.

At 71, Mathieson doesn’t have cognitive decline, and she lives independently. But she thinks about what’s ahead and worries that if she does start to lose her faculties, she could become a burden to those around her. After spending some time with a robot, she felt excited about the possibilities of a helper that might allow older people to maintain their independence a little longer. “Robots are the future,” she says. I ask her if she’d want to have a robot in her home. “I would be comfortable if there was no human available to do that same function for me. But I do think that in my generation, people would prefer human interaction, not just in health care but in all aspects of their daily lives.”

All generations age, and all of us approach this narrowing of time with uncertainty. Yet we seem to be at a defining moment: Deciding how we get old—and how we take care of each other— will reveal deeper truths about our society, its fault lines and strengths. The great hope is human care that’s augmented by automation; the great fear is replacement. Will we get to choose, as Mathieson hopes, the shape of our final years? Or have we ever?

FILED UNDER: