In a lab in one of the old buildings at the back of Trinity College, a perky-looking robot called Stevie is rolled out to meet his latest visitor. He introduces himself and tells a cheesy joke before giving a rendition of Bonnie Tyler's 'Total Eclipse of the Heart', lifting up his stubby arms in an awkward dance as he sings. It's hard not to smile watching him.
The robot standing in front of us is in fact Stevie II, a revamp of Ireland's first socially assistive robot with artificial intelligence features. Designed to support the elderly, he can interact with people using speech, body language and facial expressions, as well as carry out functional tasks like reminding users to take their medication, controlling lights and TVs, and making video calls.
He stands at about 1.4 metres tall and weighs in the region of 60 kilos. After being greeted, he comes to life with a gentle British voice - an accent preferred over other available options during early tests.
Stevie has been built from scratch in this very same room, on the top floor of Trinity's Department of Mechanical and Manufacturing Engineering. The Robotics and Innovation Lab, as it's called, is a busy space filled with worktops, machines, and boxes with bits of plastic, wood and fabric. It looks more like an art workshop than a university lab. Over a cabinet at the side of the room is a big blow-up of a photo of Stevie on a recent Time magazine cover - a testament to his growing international profile. The poster was printed for display at a college open day before being gifted to the development team.
There are six researchers in all in the group, most of whom began their studies here as undergraduates before going on to postgraduate level.
These last few months have been especially fruitful for the project, with major pilots taking place in England and the United States. The research group is funded by Enterprise Ireland and now also has its own registered company, Akara Robotics - a play on the Irish word 'cara', meaning 'friend'. ("We tweaked it for internationalisation," says one of the engineers).
Leading the initiative is Dr Conor McGinn, a young professor who talks eagerly about its social mission. "A lot of us on the team started at a time when the economy was doing pretty badly," he says. "We're the product of a generation that had to work really hard, and that's forged some character in us. Any one of our engineers could walk into a job that pays twice or three times what we pay - with an awful lot more security and creative comforts - but they don't. We've all been willing to make sacrifices."
With the lab's track record, McGinn is certainly within his rights to speak idealistically. He tells me he tends to describe Stevie as a project centred on wellbeing and enjoyment rather than technical advancement.
"Laughter isn't what usually what comes to mind when we think about retirement communities," he explains. "We think of settings like nursing homes as places without much joy; places where people don't want to go. What a piece of technology like Stevie can do is bring people together and provide a focal point for interaction and a bit of craic."
McGinn stresses that the team's work has been driven from the outset by user feedback. There is a regrettable tendency among technology developers in the field to prioritise novelty over genuine needs, he adds. "We're not the kind of engineers who build a robot and only then try to find users. We see ourselves instead as engineers looking to understand and build solutions to already existing problems.
"Technology, for us, is a tool rather than an end in itself."
Indeed, Stevie's development has been significantly adapted in recent months as a result of focus groups. "We were initially very caught up on applications," he says. "What came back to us as a real surprise is that people were generally more concerned with the experience of interaction than with what the robot was physically doing. Our focus shifted from there away from tasks like fetching and picking up stuff, and towards making people laugh.
"The main thing people wanted was to be able to have a good time, so that forced us to consider aspects of personality that we hadn't really thought through before."
While it might sound like just cheerful distraction, Stevie and robots of his kind have the potential to transform elder care as we know it.
The problems facing the sector are widely recognised, chief among them poor working conditions and high staff turnover, as well as the ever-increasing demand on services.
The ageing of Ireland's population is a trend unlikely to be reversed over the coming years, with ESRI research published two years ago predicting a growth of up to 54pc in demand for home-help hours and places in long-term care settings such as nursing homes, by 2030.
In a decade's time, according to projections, one in six of us will be 65 and over, compared to one in eight in 2015.
Where resources might fall short, Stevie steps in to plug the gap. McGinn is keen to emphasise, though, that the robot being developed here is intended to support rather than replace carers. "The people advocating this technology more than anyone else, in our experience, have actually been the workers themselves," he says. "These are people who see great value in their jobs, but aren't feeling at all supported. What we're trying to do is understand what would be helpful to them in managing the pressures of the work."
McGinn describes, in one example, how Stevie could allow staff to be freed up to focus on more individualised care during group activities. "We've found that what often happens in adult day centres or retirement communities is that one member of staff has to facilitate the activity - calling out bingo numbers, say, or quiz questions - while also looking after people who require extra help. It can be incredibly stressful to split your time between those two things. With a robot taking the lead, that staff member can not just sit with the person who needs individual support but actually engage with the activity and help drive the energy.
"It's amazing seeing the difference in engagement levels when we run the same event with and without a robot."
Stevie could also enable workers to develop new skills, according to McGinn. The team hopes staff in residential settings will eventually learn to add content of their own to the robot. When I visit him in December, Stevie is just back from a two-week trial at a Cornwall day centre called Reflections, catering mostly to people with dementia.
This was the first time Stevie was left alone with care staff, with the team in Dublin keeping an eye on developments remotely over the internet.
Jennifer Jenkins, who manages the centre, tells me she was taken aback by how well clients took to him. "The feedback was brilliant," she says. "We had some who were a little apprehensive at the start and then became more interested, but most people fell for him straight away. We even had people cheering when Stevie would come into a room."
Jenkins reckons that Stevie particularly appealed to the women who took part in the programme. "He tells people he's two years old - and he does almost come across as a child - so the motherly instinct might be a factor there," she explains.
This aligns, in fact, with the findings from American focus groups, in which female testers were more inclined to see the robot as a living thing, referring to it using words like "cute" .
At the centre in Cornwall, Stevie has understandably appeared even more human-like to those with memory problems. "I don't know if everyone realised he was a robot," says Jenkins. "People would ask if he had a brother or sister, or a girlfriend, and ask where he came from. They spoke to him as if he was a real person."
Any suggested improvements?
"We could have done with some older songs during musical bingo."
An earlier six-week pilot over the summer saw Stevie deployed to a military retirement community in Washington DC, where he had already spent some time earlier in spring. Knollwood, a non-profit facility run by the Army Distaff Foundation, offers independent and assisted living units to retired military officers and their families. McGinn and another member of the Trinity team, AI specialist Niamh Donnelly, moved into the community themselves to keep track of things, bringing a broken-apart Stevie across the Atlantic with them in boxes.
Jessica Herpst, Knollwood's deputy director of operations and technology, says the robot noticeably lifted the mood among residents in the time it spent on site. "There were people who would go to activities and sit at the back reading magazines and pretending not to be involved, but you would slowly notice them moving up closer to the front of the room. We found quieter and more reserved residents were really coming out of themselves as well."
Residents, in some cases, felt more comfortable confiding in Stevie than in others around them. "We did rounds of interviews where they knew they would be speaking only to the robot and that nothing would be stored," says Herpst. "What we found is that a lot of people opened up more to Stevie than to staff and even their friends and family."
Apprehension felt by staff also seems to have lessened as time went on. Herpst tells me that a carer in charge of bingo sessions was initially frustrated about having to include Stevie in the activity, feeling the task of handling a group of residents was challenging enough without also managing a robot. But the same staff member found the programme actually allowed her more one-on-one time with residents, leaving her with only positive things to say about Stevie by the end of his stay.
Sense of security
As well as encouraging social interaction, Stevie is able to help streamline care delivery in both private homes and institutional settings by briefing carers and nurses, for example, or calling for help during emergencies.
For Brendan Crean, a 72-year-old living alone in Kilbarrack in Dublin, Stevie provided a sense of security during a visit earlier this year. "It's good that he'd have a list of numbers of people to contact if I fell," he says.
But companionship is the main appeal for Crean, a wheelchair user with cerebral palsy. "I found him great company," he says of Stevie. "If you don't have family nearby, like me, you can always put Stevie on and have a chat and a bit of banter. You get lonely on your own with the weather being miserable, you know?"
There are challenges, however, to having Stevie operate in independent living spaces. The first and most important issue is safety, explains McGinn. "If you're in a home with one or possibly two people and something goes wrong, there probably won't be someone qualified on site to intervene. That's unlikely to be the case in a group setting."
Mobility is another consideration, since the robot needs room to move without flights of stairs or narrow doorways. And then there's the issue of cost: Stevie provides much greater value for money, of course, in environments where he can serve several people at once. Still, McGinn adds that a simpler version of Stevie could eventually be built for use in private homes. The focus in these circumstances would be on critical applications, he says.
Seán Moynihan, CEO of the charity Alone, agrees that technology has a crucial role to play in elder care. The charity has been working with McGinn and his team since 2017 to ensure their designs fit the needs of older people, facilitating feedback sessions. "We find people get excited about being involved," he says. "When they're reading the papers and watching TV, it's all about tech, and this kind of consultation really makes them feel part of that world."
Moynihan is also conscious, however, that there tends to be less informal support for pensioners today at a community level. "Some technology developers - not the Trinity team - have put out things like robots as a way of substituting for human interaction," he says. "But unless there's someone available to respond when a problem is escalated by Stevie or whatever else, it's no good."
McGinn, too, emphasises the need for properly funded social infrastructure. He pauses to reflect when I ask how feasible it would be to have something like a Stevie in every nursing home in Ireland within the next decade or so.
AI robots should only be considered as part of an integrated approach to elder care, he responds. "I think it's something that shouldn't happen in a vacuum. As good as it might be for my personal ego, technology is only part of the solution."
The problems facing the sector are only going to worsen, he says, because of changing demographics. "We're going to be in big trouble if all we have to offer are Band-Aid solutions," he adds.
Whatever it might mean for the future of the sector, Stevie will soon be part of a market that looks set to only expand over the coming years. Some of the most significant developments are taking place in Japan, where a government robot strategy is targeting an increase in the number of care recipients willingly using nursing robots to 80pc by 2020.
The plan also aims to prevent work-related backache among carers by rolling out robotic aids to lift and move elderly patients. In Tokyo's Shintomi nursing home, 20 different types of robots are said to be in operation, including Paro the seal, Aibo the dog, and a humanoid named Pepper. Demand for assistive healthcare robots is forecast to rise.
Stevie might still be a prototype, but the results from his own trials have shown big potential. The latest model is more dexterous and mobile than its predecessor as well as more socially intelligent: McGinn estimates that four times more computing work has gone into building Stevie II, which was launched this past May.
Still, McGinn acknowledges readily, many of the tasks and actions it performs are still remote controlled. "It's hard to programme when we're not yet sure what it should be doing in a given situation. We're learning anecdotally what a successful outcome means in different circumstances; what worked well and what didn't. It's common enough when people build healthcare robots to not really spend much time in the settings where they're going to be deployed. But the assumption that things can be fully figured out in the lab is a recipe for failure."
A few more months of work are needed before Stevie can be permanently deployed and manufactured on a wider scale. This final stage will involve revising almost every aspect of the design, improving safety, software and applications. If all goes to plan, a completed robot will be dispatched to Knollwood later in 2020 - the first permanent deployment after three years of work.
An unravelling of our social fabric?
McGinn is reluctant to give details on pricing, although reports suggest the robot is expected to be made available on a monthly contract undercutting the equivalent cost of human labour. While sociable robots like Stevie are unlikely to be able to entirely take over from carers, it could be argued that their very existence means we've accepted performative connectedness in circumstances where actual human connection is in short supply.
Some older people seem to view these kind of robots as friends, but of course, the sentiment can never be reciprocated. Although he might ask how we are, in other words, Stevie lacks the capacity to actually care about how we're doing.
With the mainstreaming of AI robots, as the American psychologist Sherry Turkle has asked, have we settled for the "acting out" of caring over genuine caring relationships? Are robots just another step in the unravelling of our social fabric? Or could they instead improve the delivery of care by easing the stresses faced by staff?
For all the assurances given to us that such technology will augment rather than replace human interaction, only time will tell if that will remain the case.
Changing demographic and economic conditions could result in robots fulfilling unintended briefs, after all. There have even been calls in the UK for a government guarantee that robots and remote monitoring systems will never be imposed on people or used as a substitute for caregivers.
What bodes well for the future of Stevie, though, is the fact that he's been designed around the expressed needs of older people. We can only hope that the goals of developers like McGinn and his team are what will define the long-term outcome of this robotic moment. It could turn out to be empowering for everyone, or just absurdly miserable.