Thomas would sit still all day at his aged-care home. The 82-year-old had not spoken for two years. Staff would serve him coffee and leave him be.
One day, a group of researchers led by Wendy Moyle, from Queensland's Griffith University, were running a test involving a Paro robot seal called Millie. They had been used in Japan to help children with developmental disorders, and to comfort victims of the 2011 tsunami. Professor Moyle wondered if a Paro could improve the quality of life for dementia sufferers by making them less anxious - the seals are covered in antibacterial plush fur and sensors, and squirm and squeal when stroked.
So Millie was handed to Thomas. He looked at it with surprise but put it on his shoulder and started petting it. Millie turned to him, lifted its head up, looked at him and made a noise. Squeee! Thomas looked very content. He did not want to hand her back, so they told him Millie was going to bed and to say goodbye to her. Thomas took Millie off his shoulder, looked at it and said, "Goodbye Millie" - his first words in two years. Around the room, people were in tears. They had assumed he could not speak.
"We wondered whether people with dementia would throw them out and say that's just a stuffed toy," Moyle says. "I guess we were surprised by the reaction of the group in the study - they were keen to engage with the robot, look at what it was, and some of the people who were cognitively impaired would ask questions about Millie."
And Paro's appeal is not limited to the elderly: Moyle says if she leaves one on the table in a room of academics, everyone starts playing with it. "There were a lot of people saying things like, I hadn't thought about robots before - this is so adorable. The reason why it's so adorable is because it's responsive, like a dog." You engage Paro by stroking and touching it. "It looks up at you with long eyelashes and very dark eyes and it's like having a newborn baby. It has that same emotional response."
So why do people react to something that is not alive? "I think you forget that. I find that even myself."
Researchers believe we will become emotionally attached to robots, even falling in love with them. People already love inanimate objects like cars and smartphones. Is it too far a step to think they will fall deeper for something that interacts back?
"Fantastic!" says Adrian Cheok, of Japan's Keoi University's mixed reality lab, when told of the Paro study. Professor Cheok, from Adelaide, is at the forefront of the emerging academic field of Lovotics, or love and robotics.
Cheok believes the increasing complexity of robots means they will have to understand emotion. With social robots that may be with you 24 hours a day, he says it is "very natural" people will want to feel affection for the machine. A care-giver robot will need to understand emotion to do its job, and he says it would be a simple step for the robot to express emotion. "Within a matter of years we're going to have robots which will effectively be able to detect emotion and display it, and also learn from their environment," he says.
The rather spooky breakthrough came when artificial intelligence researchers realised they did not need to create artificial life. All they needed to do was mimic life, which makes mirror neurons - the basis of empathy - fire in the brain. "If you have a robot cat or robot human and it looks happy or sad, mirror neurons will be triggered at the subconscious level, and at that level we don't know if the object is alive or not, we can still feel empathy," Cheok says. "We can't really tell the difference if the robot is really feeling the emotion or not and ultimately it doesn't matter. Even for humans we don't know whether a person's happy or sad." He argues if a robot emulates life, for all intents and purposes it is alive.
Psychologist Amanda Gordon, an adjunct associate professor at the University of Canberra, is sceptical. "It's not emotional, it's evoking the emotion in the receiver,'' she says. ''That seal isn't feeling anything. It's not happy or sad or pleased to see you."
She says the risk is that people fall for computer programs instead of a real relationship. "Then you're limiting yourself. You're not really interacting with another. Real-life relationships are growth-ful, you develop in response to them. They challenge you to do things differently."
Cheok's research shows 60 per cent of people could love a robot. "I think people fundamentally have a desire, a need to be loved, or at least cared for," he says. "I think it's so strong that we can probably suspend belief to have a loving relationship with a robot."
Probably the most advanced android in the world is the Geminoid robot clone of its creator Hiroshi Ishiguro, director of the Intelligent Robotics lab at Osaka University. Professor Ishiguro says our bodies are always moving, so he programmed that realistic motion into his creation along with natural facial expressions.
The one thing it does not do is age, which means 49-year-old Ishiguro is constantly confronted with his 41-year-old face. "I'm getting old and the android doesn't,'' he says. ''People are always watching the android and that means the android has my identity." So he has had plastic surgery - at $10,000, he says it is cheaper than $30,000 to build a new head.
Robots can help kids with autism who do not relate to humans. Ishiguro is working with the Danish government to see how his Telenoid robots can aid the elderly.
Moyle says she has had inquiries from throughout Australia about Paro. A New Zealand study showed dementia victims interacted with a Paro more than a living dog.
"There are a lot of possible negative things [that artificial intelligence and robots could lead to]," Cheok says, "and we should be wary as we move along. We have to make sure we try to adjust. But in general I think the virtual love for the characters in your phone or screen or soon robots is ultimately increasing human happiness, and that's a good thing for humanity."
Sign up for our newsletter to stay up to date.