Making friends with robots
How should we properly address, and relate to, the digital humans increasingly popping up in our everyday lives? Katie Kenny reports.
We were trying to get in touch with our internet service provider. I can’t remember the reason. But we contacted the company through its website chat system.
My partner was typing, and I noticed his language was unusually clipped, devoid of the words ‘‘please’’ or ‘‘thank you’’.
‘‘You don’t have to be so rude,’’ I said.
‘‘It’s just a bot,’’ he replied, shrugging.
I took a closer look at the conversation. ‘‘No, that’s not a bot. You’re talking to a real person.’’
The speed, and humanity, of the responses was beyond the capabilities of common virtual assistants, I thought.
So we asked the woman at the other end about the weather, and what she did at the weekend. Her replies confirmed my suspicion. She sounded like a person used to being hassled in this online role, wanting to end the conversation quickly.
Our uncertainty felt like a very 2019 thing. Voice-controlled artificial intelligence systems, and even robots, have become more common in our everyday lives; from Siri, Apple’s ‘‘intelligent personal assistant’’, to WoeBot, the chatbot therapist, to Travelmate, the suitcase that uses GPS to stay close to your connected smartphone.
As they proliferate, how should we properly address, and relate to, these virtual beings?
IBM distinguished designer Adam Cutler, at AI-Day in Auckland last year, said society was shifting from ‘‘a transactional age of computing, to a relationship age’’.
Eventually, people will want to date their AI operating systems, he said, alluding to Spike Jonze’s 2013 film, Her, about a man who falls in love with his operating system.
‘‘Why? Pathetic fallacy. We, as humans, want to attribute human feelings to inanimate objects. We want to form relationships.’’
In his TED talk, Cutler adds: ‘‘For the past 72 years, we’ve been communicating with computers on their terms. All of the user interfaces we’re surrounded by are nothing more than elaborate workarounds for us to share our intent with a computer.
‘‘Today, we’re right on the cusp of an evolution in our relationships with humans and machines. These machines aren’t programmed, they’re taught. This means a machine can understand, reason, learn and interact, and these are the very building blocks of what a machine needs to form and maintain a relationship with a human.’’
One way to foster that relationship is for the AI to look, well, human, says Greg Cross, chief business officer at Soul Machines.
‘‘With the technology that’s being developed, we’re going to spend more time interacting with machines. At Soul Machines we’ve got a simple vision: aren’t machines going to be more helpful to us, if they’re more like us?’’
The Auckland-based company is known around the world for its creation of ‘‘digital humans’’ — autonomous, animated individuals that look and sound like real people, powered by virtual central nervous systems.
These digital humans have been employed at banks, airlines, education and healthcare services.
‘‘We believe by adding a face to AI, we’re actually allowing large organisations to provide a much more personalised customer experience,’’ Cross says. Pilots using digital humans at NatWest branches in the United Kingdom and at Air New Zealand showed consumers were ‘‘quite happy’’ to interact and even form emotional relationships with them.
However, he adds, the aim of digital humans isn’t to replace traditional customer service staff. ‘‘The simple reality is there will always be customers who have problems which are very complex, and having resources available to provide real human interaction will be required as well.’’
When it comes to robotic companions at home, however, those who grew up watching The Jetsons might imagine something more like Rosey the Robot.
First produced in 1962, the futuristic cartoon series was, reportedly, set in 2062. But Hanns Tappeiner, president and co-founder of home robotics company Anki, says Rosey could become a reality within the next couple of decades.
The company’s latest, and most advanced, creation is called Vector. A pocket-sized robot, rather reminiscent of Pixar’s WALL-E, its goal is to become ‘‘part of the family’’. Autonomous and always on, Vector reacts to people and things around it.
At first, Vector and I struggled to understand each other, given my New Zealand accent and his preference for Americans.
But once he had registered my voice commands, he’d respond with beeps and whirrs, the weather forecast, or the answer to a question I’d asked him to look up. (Cloud-controlled, he self-updates.)
He can take photos, set a timer, give me a
fist bump, and return to his charging station when his battery is low. He’s a combination of a home assistant, and a pet. (More pet, at this stage, I’d say.)
On the 18th floor of Anki’s headquarters in San Francisco, a wall is decorated with lists of animals, and their properties. From parrots, to sugar gliders, to golden retrievers. Vector is ‘‘earnest’’ and ‘‘bright’’, like a border collie. He’s ‘‘watchful’’ and ‘‘skittish’’, like Bambi.
Most of the people working on this floor belong to Anki’s inhouse character studio.
‘‘The character and personality of Vector is at least as important as the technology itself,’’ Tappeiner says. ‘‘We’re essentially trying to create a real-life version of a robot you’d only normally see in a movie.
‘‘Early on, we thought, how tough can it be? It can be very tough. Engineers don’t necessarily make good character developers.’’
Anki’s 20-strong character team – made up mostly of former DreamWorks and Pixar animators and character directors – is one of the company’s fastest growing.
Personality, Tappeiner explains, completely changes how people interact with Anki’s robots. Increasing the frequency and duration of eye contact, for example, boosted the length of time people spent interacting with
Vector.
‘‘We’re trying to reverse engineer human behaviour by looking at how people react to our robots. And that’s superimportant for the next products we’re making.’’
The company’s ultimate goal is to get a robot into every home. (About 1.5 billion of its robots are in homes around the world now.)
While home assistants such as Alexa could, broadly, be considered robots, they’re more like ‘‘pieces of hardware which incorporate a fair amount of intelligence,’’ Tappeiner says.
His ideal home robot is more like Rosey; something that can clean up, wash dishes, and fold laundry. ‘‘That sounds crazy because it’s fairly far off but . . . I’m 100 per cent sure it’s going to happen. The question is just when, and who’s going to be there first.’’
While Tappeiner says we’re about 15 years away from this ideal, give or take five years, others think we’re much further.
Professor Malcolm MacIver, of the Neuroscience and Robotics Laboratory at Chicago’s Northwestern University, studies ways that complex animal behaviour can be applied to making advanced biorobotic systems.
Tasks that can seem simple for humans are actually really hard for robots, he says. ‘‘I’ve been working for a long time on cracking the sensory motor crux of some of the challenges we face with making the robot we all want to have, which is, of course, the one that can fold laundry and get groceries,’’ he tells me during his trip to New Zealand, for the International Science Festival in Dunedin.
Humans and other animals are good at taking in lots of sensory information, and generating movement appropriate to that information. (Think about how you’d grasp a coffee cup differently, depending on whether the cup is made of ceramic or paper, for example.)
‘‘Believe it or not, we’re quite a ways from getting a robot to do that kind of manipulation,’’ MacIver says.
How far, exactly? He sighs. ‘‘We’re, I would say, further behind than people think.’’
Because tasks such as picking up a cup of coffee, or even folding clothing, are so routine for humans, we tend to underestimate how difficult they’ll be for machines, he says.
In the near future, it’s more likely we’ll develop meaningful bonds with AI systems, rather than Rosey-like robots. ‘‘It’s possible we’ll develop a relationship with something like an AI agent on our phone. That’s more AI than robotics. It’s a blurry line.’’
On top of technical difficulties, there are social challenges. Even if one day we do work and live alongside intelligent robots, what strengths and weaknesses, opportunities and threats, do they present? If they can work all day, every day, will we have to tax them?
It’s easy to become ‘‘freaked out by The Terminator’’ and forget about the positive things technology can offer humans, says Sarah Hindle, general manager at Auckland’s Tech Futures Lab.
While today’s digital humans can have ‘‘somewhat of a conversation’’, she says, they’re still a long way from having human-like relationships. But that doesn’t mean they can’t help combat, say, loneliness.
‘‘Imagine being able to get healthcare access and
information to remote locations through a digital human, or being able to give personal tutoring to people in those places,’’ she says.
Research suggests patients are more willing to disclose personal information in a clinical setting to virtual humans, rather than actual ones.
Perhaps a similar mentality explains why my partner felt no need to be polite to the supposed chatbot. I later called the company, to settle our debate over whether the sales assistant was a human. ‘‘Sometimes it’s a bot and sometimes it’s a real person,’’ the woman said.
‘‘We’re trialling a bot, but when a customer is talking to it there will be a function that allows them to switch to talking to a real person.’’
‘‘So, I’d know if I were talking to a bot?’’
‘‘Yes,’’ she said. ‘‘You’d know.’’