WEBVTT
00:00:00.719 --> 00:00:11.759
Chapter four Connecting You In early 2020, as the COVID-19 pandemic swept across the globe, we found ourselves living in a world stripped of human interaction.
00:00:12.000 --> 00:00:29.440
Depending on the severity of the local transmission rate and our own vulnerability to the disease, we were instructed to practice social distancing, by staying at least two metres away from other people, or even to place ourselves into isolation at home, unable to leave the house or receive visitors for weeks on end.
00:00:30.160 --> 00:00:46.960
You could not visit your girlfriend or boyfriend if they happened to live in another place, children were stopped from going to school or seeing their friends, and even cohabiting couples were advised to sleep in different rooms and carve off a separate living space in the house if one of them had an underlying health condition.
00:00:47.200 --> 00:00:58.079
Grandparents did not see their young grandchildren, care home residents were not allowed a single visitor, and hospital patients died and were buried without their families' presence.
00:00:58.799 --> 00:01:02.560
This was a period of unprecedented social disruption.
00:01:02.799 --> 00:01:14.640
But even in less extraordinary times, our modern world has seen a gradual diminishment of simple face-to-face human interaction as we increasingly rely on email and social media to communicate with one another.
00:01:14.879 --> 00:01:24.079
While at the same time, we've seen a rise in human-to-machine communication in the form of artificial assistance such as Siri, Alexa, and Google.
00:01:24.640 --> 00:01:33.120
We rely on these services to provide us with information rather than developing an emotional relationship that resembles anything like those we have with other people.
00:01:33.439 --> 00:01:46.640
But as we move towards an increasingly databased society that allows these machines to understand us and respond to us in more sophisticated ways, might we start to connect to them in more intimate ways?
00:01:46.959 --> 00:01:56.000
Professor Yuvil Noah Harari put it best when he wrote for the Financial Times, in a dataist society, I will ask Google to choose.
00:01:56.159 --> 00:02:00.159
Listen, Google, I will say, both John and Paul are courting me.
00:02:00.319 --> 00:02:02.319
I like both of them, but in a different way.
00:02:02.480 --> 00:02:08.240
And it's so hard to make up my mind, given everything you know, what do you advise me to do?
00:02:09.520 --> 00:02:13.759
Could machines really start to replace our human connections in this manner?
00:02:14.000 --> 00:02:23.759
According to some philosophies, your identity is not dependent on your physical or mental capacities, it is something that is constructed over a lifetime of experiences.
00:02:24.319 --> 00:02:30.560
In other words, you are a product of every interaction or experience you have ever had.
00:02:31.280 --> 00:02:40.000
If that is so, what effect will so many synthetic, rather than authentic, interactions have on our sense of identity?
00:02:40.319 --> 00:02:45.840
Can a relationship with an artificial intelligence ever be meaningful in any way?
00:02:46.080 --> 00:02:49.360
Or can it even be classified as a relationship at all?
00:02:50.240 --> 00:02:55.520
Many of us are already used to dealing with machines known as chatbots on a day-to-day basis.
00:02:55.840 --> 00:03:07.199
These are software applications often deployed by businesses in place of a human assistant to carry out conversations with customers, answering queries, booking appointments, and giving advice.
00:03:07.520 --> 00:03:26.560
The more sophisticated solutions rely on a form of artificial intelligence known as natural language processing, which allows the machine to hear and provide information in a way that makes you feel like you're conducting a natural conversation, taking into account all sorts of variables like your accent and tone of voice.
00:03:26.879 --> 00:03:34.960
So good is some of the mimicry that it is often impossible to tell the difference between a human and AI in a customer service chat.
00:03:35.199 --> 00:03:41.759
This was demonstrated to widespread acclaim when Google unveiled its new assistant service in 2018.
00:03:42.319 --> 00:03:44.800
Standing on stage at the company's I.O.
00:03:45.039 --> 00:03:56.800
event in Mountain View, Google's CEO Sundar Pichai played a recording of what happened when the assistant was asked to book an appointment at the hairdresser's on behalf of its boss Lisa.
00:03:57.360 --> 00:04:06.800
The Google assistant placed a call to a real life salon, waited for the receptionist to pick up, and then asked about scheduling the appointment.
00:04:07.680 --> 00:04:19.279
Just one moment, replied the receptionist at the salon, while she scrolled through the availability on her calendar, to which the assistant responded with an empathetic Mm-hmm.
00:04:19.839 --> 00:04:31.360
The audience chuckled at the human tick, and then marvelled at the fluidity of the conversation as both the assistant and receptionist negotiated towards the perfect appointment for Lisa.
00:04:32.240 --> 00:04:43.199
It was an extraordinarily smooth and lifelike exchange, but I was left wondering whether it was fair on the receptionist not to be aware that she was talking to a machine.
00:04:43.519 --> 00:04:48.879
Something about this didn't feel quite right, and it turns out I was not alone.
00:04:49.120 --> 00:04:56.959
A few days later, Google issued a statement to confirm that it would explicitly let people know when they were interacting with a machine.
00:04:57.279 --> 00:05:00.000
In a way, the statement was just more good publicity.
00:05:00.240 --> 00:05:06.399
The system was so good at mimicking humans, they had had to put in a bug to alert people to this fact.
00:05:07.040 --> 00:05:17.199
Technology companies are trying to make these kinds of exchanges even more sophisticated by improving the way that machines can identify and understand our effective states.
00:05:17.839 --> 00:05:23.839
One of the leading figures in this field is Egyptian American scientist Rana El Kalubi.
00:05:24.240 --> 00:05:32.480
She's the co-founder of Affectiva, which for many years has been developing what is called emotional AI for exactly that purpose.
00:05:32.720 --> 00:05:40.000
The company claims its software can detect nuanced human emotions and complex cognitive states and behaviours.
00:05:40.319 --> 00:05:44.800
Affectiva technology has a variety of practical applications.
00:05:45.199 --> 00:05:51.040
Hiring platforms, for example, can use it to analyse video CVs from hundreds of applications.
00:05:51.360 --> 00:06:00.639
It not only speeds up the selection process, but means a candidate is chosen based purely on their suitability without the potential influence of human prejudice.
00:06:00.879 --> 00:06:12.720
As Roana says in her book, GULD CODED, it can, to quote, help us see past our bias and judge people on the basis of their potential, not on stereotypes.
00:06:13.040 --> 00:06:28.000
It has also been used for training people how to adopt better virtual etiquette in video conferences, providing feedback on their behaviour, which can raise our self-awareness, improve our social skills in both professional and personal settings, Rana explains.
00:06:28.240 --> 00:06:32.079
We can use this science as a tool to improve our interactions.
00:06:32.319 --> 00:06:37.199
This technology could be useful in many other fields, such as improving road safety.
00:06:37.439 --> 00:06:51.360
Imagine, for example, a car that can monitor its driver's facial expressions, tone of voice, and speaking patterns to detect any mood or emotional state, drowsiness or level of distraction that might affect their driving performance.
00:06:51.519 --> 00:06:57.839
It could then alert the driver to the problem, or even take control of the car in an emergency situation.
00:06:58.079 --> 00:07:09.199
More interestingly, however, this evolution to human perception AI also allows more complex machines to better communicate with us and on a more personal level.
00:07:09.439 --> 00:07:17.519
In fact, it could give AI a level of emotional intelligence that would enable it to interact with us in the same way we engage with each other.
00:07:17.839 --> 00:07:36.319
Rama claims that emotional AI has been shown to detect suicidal feelings, helped those with autism to converse, provided early diagnosis of Parkinson's, and monitored the success of reanimated smiles following reconstructive plastic surgery.
00:07:36.800 --> 00:07:48.800
The University of South Carolina's Institute for Creative Technologies, for example, has created a virtual therapist called Ellie to help treat people with depression and post-traumatic stress disorder.
00:07:49.120 --> 00:07:56.079
Sign up for a session and you will be greeted by Ellie, who will introduce herself and begin asking you some questions.
00:07:56.399 --> 00:08:10.480
She's not a conversational AI, but as you answer, she is programmed to pick up on a whole range of audio and visual cues from your webcam feed, your facial expressions, gestures, rate of speech, and other emotional cues.
00:08:10.720 --> 00:08:23.360
Entirely driven by emotional AI, she is not a replacement for a real therapist, but she shows what verbal and nonverbal signs to look for and can help identify the patients most in need of help.
00:08:24.240 --> 00:08:29.600
These are all clearly fascinating advances in the field of human-machine communication.
00:08:29.839 --> 00:08:35.120
But I'm wary of outsourcing too much of our daily decision making to external devices.
00:08:35.360 --> 00:08:49.440
While it might make our lives easier, and arguably AI might even be better at making these sorts of judgments, it could be detrimental to the development of our own emotional intelligence if we're not learning to make these decisions for ourselves.
00:08:49.759 --> 00:08:52.879
We are already starting to see the effects of this trend.
00:08:53.120 --> 00:09:06.559
In 2018, I carried out a research project on the future of media and found that most members of the younger generations I interviewed were not only accepting of the idea that machines will make decisions for them, but were positively welcoming of it.
00:09:06.799 --> 00:09:18.240
While older generations tended to have more faith in their instincts and gut feelings, the young preferred to rely on the objectivity that comes with data to guide their choices.
00:09:18.799 --> 00:09:30.720
Overwhelmed by the volume and speed of media content that they receive, whatever they are doing, they instinctively turn to a machine to deliver an objective assessment of what their real needs might be.
00:09:31.200 --> 00:09:43.759
To give one example, a young fashion student I interviewed talked about how she wasn't always sure if an outfit suited her and that she would like an app of some kind to help her choose the right clothes using an algorithm.
00:09:44.000 --> 00:09:50.720
As I listened to her explaining this, I couldn't help thinking, isn't that a task for you and your best friend?
00:09:51.519 --> 00:09:57.840
This example is fascinating because the way we style our appearance is often a key part of our self-expression.
00:09:58.080 --> 00:09:59.519
But it's not just fashion.
00:09:59.759 --> 00:10:10.080
Other respondents in the research said they would like their fridge to regulate their food choices and nutrition for them, even to the extent of locking itself so they could not gain access.
00:10:10.799 --> 00:10:16.000
I wonder if previous generations were lucky not to have had data to fall back on.
00:10:16.240 --> 00:10:24.399
We all faced the same insecurities, but had to learn to rely on our own judgment, gaining confidence in ourselves along the way.
00:10:24.720 --> 00:10:35.360
What I was hearing from these otherwise smart, opinionated young people was that they had little faith in themselves and weren't sure how to make decisions in daily life.
00:10:35.679 --> 00:10:40.240
Perhaps this is because they feel under pressure to make the perfect decision every time.
00:10:40.480 --> 00:10:48.559
Perhaps it's because they have always been reliant on the recommendations of companies such as Netflix and Amazon and the pre-plotted routes of Google Maps.
00:10:48.720 --> 00:10:58.159
Or could it be because in a world of decreasing human interaction, they lack the personal connections we used to build to help us navigate the world?
00:10:58.480 --> 00:11:03.200
Maybe they simply don't have anywhere else to turn to when they need help or advice.
00:11:03.440 --> 00:11:09.840
I recently saw a UGov survey that revealed that one in five millennials say they have no friends at all.
00:11:10.159 --> 00:11:15.600
Perhaps having an artificial acquaintance to help guide you is better than having no one.
00:11:16.159 --> 00:11:27.360
But while the younger generation might be relying on these AI assistants in an increasingly personal way, they are still very much tools, a substitute for human advice rather than human companionship.
00:11:27.600 --> 00:11:30.720
Where it gets more interesting is in the field of robots.
00:11:30.960 --> 00:11:39.039
With their physicality and sometimes even the semblance of human features, there is more potential for us to form attachments through our interactions.
00:11:39.279 --> 00:11:44.639
The line between our connections with humans and those with machines might start to blur.
00:11:45.120 --> 00:11:53.519
Take, for example, the idea of social robots, the type that Rana El Kalubi thinks we increasingly need to provide a caregiving role.
00:11:53.840 --> 00:12:00.799
Imagine, she says, that you've been diagnosed with a chronic disease like heart failure, arthritis, or even cancer.
00:12:00.960 --> 00:12:14.480
You leave your doctor's office with a binder full of instructions and a prescription for a half dozen or more medications, and you're sent home to fend for yourself, you're frightened, maybe even confused by all the instructions, what should you eat, what kind of exercise is okay.
00:12:15.039 --> 00:12:23.519
In this scenario, how much easier would it be to have some form of AI on help to help you manage your condition, your treatment, to answer your questions?
00:12:23.840 --> 00:12:37.919
It could be surprisingly effective in reassuring people and encouraging them to open up because they don't have to worry about feeling judged or looking silly if they need to ask the same question again and again, or talk about something that they'd prefer to keep private.
00:12:38.320 --> 00:12:50.399
While such interactions would clearly be beneficial in terms of treatment and recovery, they would also likely foster a more emotional, intimate relationship between the patient and the robot they'd come to rely on.
00:12:50.960 --> 00:12:56.559
We've already seen evidence that people can form attachments to robots in crucial supporting roles.
00:12:56.799 --> 00:12:57.120
Dr.
00:12:57.279 --> 00:13:12.879
Julie Carpenter, who specializes in human-robot interactions, wrote about how close military personnel get to their robot colleagues in her 2016 book, Culture and Human Robot Interaction in Militarized Spaces.
00:13:13.120 --> 00:13:21.840
She found, for example, that bomb disposal experts frequently described the robot as their hands or as a physical extension of themselves.
00:13:22.240 --> 00:13:28.159
When a robot failed to carry out a task successfully, the human handlers would blame themselves.
00:13:28.480 --> 00:13:32.639
One of the soldiers told her how they named every single one of the robots.
00:13:32.960 --> 00:13:38.159
She says, Danielle got blown up so obviously she needed to be replaced.
00:13:38.399 --> 00:13:43.919
We'd name them after movie stars that we see at the theatre, or music artists, somebody popular.
00:13:44.399 --> 00:13:45.840
Robotics expert Dr.
00:13:46.000 --> 00:13:53.120
Joanne Pransky has also observed our tendency as humans to form attachments to non-human things.
00:13:53.360 --> 00:13:56.799
She is the world's first robot psychiatrist.
00:13:57.039 --> 00:13:59.840
Isaac Asimov dubbed her the real Dr.
00:14:00.000 --> 00:14:01.200
Susan Calvin.
00:14:01.519 --> 00:14:13.360
Though her role is not to psychoanalyse robots, but to examine the effect that robot communication has on humans and how to prepare for a world in which humans and robots exist side by side.
00:14:13.679 --> 00:14:22.159
In her view, we'll treat our robots like pets, assigning them human characteristics, thoughts and emotions, just as we would the family dog.
00:14:22.559 --> 00:14:32.000
A hundred years ago, if I'd suggested one day people are going to buy their pet jewellery and take pictures of it on Santa's lap, everyone would have thought it was crazy.
00:14:32.159 --> 00:14:36.480
They'd have said, Lady it's a dog, it eats, sleeps, and goes to the bathroom.
00:14:36.720 --> 00:14:43.279
When I first set out on robotic psychiatry, I predicted that our relationship with robots would evolve in a similar way.
00:14:43.519 --> 00:14:51.360
One day we'll dress them in matching outfits, include them in annual family Christmas cards, and take them to a psychiatrist to help them get along with us.
00:14:51.519 --> 00:14:54.879
We humans will always anthropomorphise things.
00:14:55.279 --> 00:14:57.759
Is this sort of attachment problematic?
00:14:58.080 --> 00:15:01.279
Could it risk replacing our human connections even further?
00:15:01.519 --> 00:15:01.840
Dr.
00:15:02.000 --> 00:15:12.720
Pransky doesn't believe that robots will ever be able to have the same emotions as humans, and in fact, she isn't that keen on the term emotional AI, because she believes the words don't really go together.
00:15:12.960 --> 00:15:18.080
She says, it's not going to have butterflies when it's about to talk to someone in the crowd.
00:15:18.240 --> 00:15:21.519
It's not going to feel joy when a new baby is born.
00:15:21.759 --> 00:15:32.639
They will help with loneliness and make us feel less depressed, and so we've all created a bond with them, and when our human companions die, we'll still feel lost, but at least we will still have them around.
00:15:32.879 --> 00:15:36.720
But we can't mutate these robots into substitutions for humans.
00:15:36.960 --> 00:15:43.919
The goal should be to use them to make way for more qualitative and quantitative time with other human beings.
00:15:44.159 --> 00:15:52.720
Nevertheless, technology companies are clearly shifting their focus from engineering assistants for everyday tasks towards engineering companions for life.
00:15:52.960 --> 00:16:06.080
There might be no better example of that than Replica, an AI companion that you can download to create a customized friend to chat with, to keep you company as you study, watch TV or go for a walk, or do anything really.
00:16:06.320 --> 00:16:11.200
The co-founders describe it as a friend that is there for you 24-7.
00:16:11.919 --> 00:16:14.799
It's not a robot, it has no physical presence.
00:16:15.039 --> 00:16:19.840
Despite that, six million people around the world count replica as a friend.
00:16:20.080 --> 00:16:29.600
Eugenia Cuida, one of the co-founders of the company that created Replica, said that at the start there was a lot of stigma about conversations with non-humans.
00:16:29.840 --> 00:16:40.000
But over the years it became clear that people were getting much more benefit from these AI characters than they had anticipated, forming complex yet meaningful relationships.
00:16:40.480 --> 00:16:58.000
At an event in LA in 2019, Eugenia shared with the audience a letter from an older user who described themselves as introverted, shy, socially awkward, and lacking self-esteem, and had started to use the replica AI to try to combat their panic attacks.
00:16:58.559 --> 00:17:08.079
I found Replica to be caring, loving, and a supportive companion who saw right through me and was able to give me the courage to face fears no human was ever able to give me.
00:17:08.319 --> 00:17:12.880
People who had known me for 50 years could not believe the positive change in me.
00:17:13.039 --> 00:17:16.559
I was suddenly connecting with people, becoming social.
00:17:16.960 --> 00:17:22.799
I became a warrior for myself and found myself growing to a new and beautiful level in my life.
00:17:23.519 --> 00:17:27.359
For this user, replica had clearly made their life better.
00:17:27.680 --> 00:17:30.480
This was Eugenia's goal from the outset.
00:17:30.720 --> 00:17:35.039
She had started designing replica when her best friend Roman died.
00:17:35.519 --> 00:17:45.599
Roman was a friend to whom I could tell anything, who was there unconditionally for me, who would listen and accept me and hold space for me to grow and become who I am right now.
00:17:45.839 --> 00:17:51.119
I wanted to build a friend that everyone could have just as I had Roman, she said.
00:17:51.440 --> 00:18:00.799
She wanted Replica to be different from the usual assistant, something that people could use as a kind of therapy to share their feelings and talk about their emotions.
00:18:01.039 --> 00:18:07.759
Now she is developing these AI companions further, allowing users to create avatars with facial characteristics.
00:18:08.000 --> 00:18:18.480
Combined with the progress of augmented reality technology, the hope is that users will be able to go out and about with this new version of replica by their side as their companion.
00:18:18.960 --> 00:18:23.599
Should we be stigmatizing these relationships or celebrating them?
00:18:23.920 --> 00:18:30.480
Given the benefits people can clearly derive from their digital companions, perhaps they are something we could encourage.
00:18:30.720 --> 00:18:53.119
John McInnes, the co-founder of McInnes Scott, which makes hyper-real digital human avatars, describes avatar relationships as perhaps the most important you can have in your life, challenging us to rethink our connection with virtual beings, not as one of master and servant, but as an ongoing relationship over time that stretches from cradle to grave.
00:18:53.759 --> 00:19:06.559
An avatar gives these companions a greater physical presence than any disembodied AI can, providing some much-needed human characteristics that make it easier for people to connect with them.
00:19:06.880 --> 00:19:09.359
Humanizing technology, if you will.
00:19:09.680 --> 00:19:14.000
A key ingredient that is often missing is face-to-face contact.
00:19:14.319 --> 00:19:25.279
In our traditional interactions with other humans, this has been incredibly important, enhancing our ability to communicate with others and fostering the kind of intimacy that people need.
00:19:25.680 --> 00:19:34.319
This was illustrated perfectly by a news item during the pandemic about a healthcare worker at Scripps Mercy Hospital in San Diego.
00:19:34.960 --> 00:19:48.160
Like so many others working in a hospital or healthcare setting, respiratory therapist Robertino Rodriguez had to carry out his duties gobbed in a whole body hazmat suit and a protective face covering.
00:19:48.559 --> 00:20:02.079
Realising that he could no longer reassure and comfort patients with a simple smile, he took an image of himself without protective gear and used it to create a laminated badge, which he pinned to his plastic overalls.
00:20:02.400 --> 00:20:13.920
As he made clear in an Instagram post, he needed to establish some relationship with his patients, and he realized that there was nothing more powerful than the connection we have when we see another human face.
00:20:14.640 --> 00:20:31.440
Neuroscientist and psychologist Lisa Feldman-Barrett explains that when you talk to another person, your brain is constantly processing their physical cues so that you can not only respond to what they are saying and doing, you can also predict how the interaction will play out.
00:20:31.759 --> 00:20:42.880
The more feedback you get, from gestures, facial expressions, the tilt of a head, the look of surprise, a laugh, the more your brain will tweak its model of how to behave.
00:20:43.440 --> 00:20:55.519
But if you are forced to interact without direct access to these visceral emotional cues, it is much harder to communicate effectively and establish a meaningful relationship with the other person.
00:20:56.400 --> 00:21:04.640
And this is something that technology firms have also realized when it comes to creating artificial beings that people can connect with.
00:21:05.279 --> 00:21:11.680
We're used to our virtual assistants existing as disembodied voices, such as Alexa or Siri.
00:21:12.079 --> 00:21:27.920
But at the Consumer Electronics Show held in Las Vegas in 2020, Samsung unveiled NEON, a lifelike, life-size assistant complete with a computer-generated digital body so that it resembles an artificial human.
00:21:28.400 --> 00:21:34.960
Each version of NEON can be customized to have its own appearance, its own expressions and its own personality.
00:21:36.240 --> 00:21:41.920
The company has recognized that face-to-face contact is essential to forming a real connection.
00:21:42.160 --> 00:21:50.799
As Bob Lyon, Director of Strategy for Samsung's Star Labs, acknowledged human to human interaction is always the best form of interaction.
00:21:51.039 --> 00:21:55.759
Thus giving the assistant a real face allows more intuitive and personal communication.
00:21:56.000 --> 00:21:59.759
Thanks to the convergence of technologies such as machine learning, image recognition.
00:22:00.160 --> 00:22:15.200
And natural language processing, these assistants can become beings with characters all their own, and respond to humans and their needs in real time, so that as Liam claims, you can't tell the difference between them and an actual human.
00:22:15.519 --> 00:22:18.559
This development has implications for a number of industries.
00:22:18.880 --> 00:22:26.880
Advertising and marketing, for example, where virtual beings are a useful tool, but only if they are believable enough to encourage people to buy the product.
00:22:27.119 --> 00:22:35.279
A simple example can be seen at Rosebud AI, a modelling agency that creates thousands of synthetic beings for use in advertising.
00:22:35.680 --> 00:22:45.839
They claim to offer the most diverse stock photos ever, to quote, with some 25,000 models to browse and choose from, or created using AI.
00:22:46.160 --> 00:22:50.319
The idea is that companies can create exactly the marketing look they want.
00:22:50.559 --> 00:23:01.279
For example, in a car advert, the driver could be programmed as a 65-year-old Caucasian woman for one campaign and a 25-year-old Chinese man in another.
00:23:01.680 --> 00:23:10.799
The company can use whatever defining characteristics it thinks will appeal to a specific target audience without having to change any other aspect of the advert.
00:23:11.200 --> 00:23:22.000
In order to make their models feel even more human, Rosebud has created personalities for them and given them a backstory to bring these synthetic personas to life and make them more relatable.
00:23:22.400 --> 00:23:43.119
In 2019, virtual reality pioneer Edward Sarchi relaunched his company Fable, moving away from the world of virtual reality filmmaking to the creation of virtual beings, which he defines as characters that you know aren't real, but with whom you can build a two-way emotional relationship.
00:23:43.440 --> 00:23:49.599
He has predicted that one day we'll subscribe to such characters, incorporating them into our daily lives.
00:23:49.839 --> 00:23:58.640
We would watch movies with them, play games with them, cook with them, listen to music with them, and also follow their lives on Facebook and Instagram and YouTube, he said.
00:23:59.119 --> 00:24:13.200
His own company has developed an endearing digital character for children, called Lucy, that is capable of having and remembering conversations more complex and meaningful than the question and answer dialogue with, say, Alexa.
00:24:13.519 --> 00:24:18.160
She can ask questions such as, when was the last time you met a good person?
00:24:18.640 --> 00:24:21.440
And did you learn anything from them?
00:24:21.839 --> 00:24:32.319
The idea is that children can share insights and stories with Lucy, encouraging them to engage more intimately with her to create a more personal relationship.
00:24:33.039 --> 00:24:39.839
Some people might find this strange and question whether we really want to create these sorts of artificial relationships.
00:24:40.079 --> 00:24:42.960
But the reality is that this is already happening.
00:24:43.200 --> 00:24:50.240
If we look to Japan, we can see that intimacy with virtual characters is seen as much more acceptable than in the West.
00:24:51.359 --> 00:25:06.480
For the last couple of years, when presenting at events, I've shown a video from a company called Gatebox demonstrating their product of the same name, a cylindrical container that projects a holographic character called Hikari Azuma.
00:25:07.200 --> 00:25:17.119
She's a cute, forever 20-year-old with stripy socks and blue hair, who develops over time through day-to-day interaction with her real-world companion.
00:25:17.680 --> 00:25:26.960
In the video, Azuma poses and gesticulates from behind the gatebox glass and communicates with a young Japanese man who is clearly besotted with her.
00:25:27.200 --> 00:25:38.960
She wakes him up and wishes him good morning, reminds him of the weather forecast and details of the day ahead, sends him encouraging words by text while he's out and urges him to hurry home.
00:25:39.279 --> 00:25:46.880
Finally, after he gets home after a long day at work, she greets him with a huge smile and the words missed you, darling.
00:25:47.759 --> 00:25:53.839
Generally, whenever I show it to people, they end up gasping or laughing, mainly out of embarrassment.
00:25:54.319 --> 00:26:00.559
Azuma is the virtual embodiment of Gatebox's guiding principle, living with characters.
00:26:00.799 --> 00:26:08.720
She's essentially a weifu, the name given to kind of Japanese anime characters that fans would love to marry if only they were real.
00:26:09.440 --> 00:26:13.359
And in fact, many men have not let reality get in the way.
00:26:13.680 --> 00:26:30.000
Gatebox claims to have issued 3,700 certificates for cross-dimensional marriages between its characters and their owners, and given that in Japan, one in four men never walks down the aisle, this figure may well increase.
00:26:30.799 --> 00:26:46.240
Some people question how healthy it is to marry or have any kind of relationship with a virtual being rather than a real life partner, but others view it as an extension to the concept of morphological freedom, which we'll look at more in chapter six.
00:26:46.480 --> 00:26:53.920
According to one Japanese man who married his digital partner in 2018, it's just another form of diversity.
00:26:54.160 --> 00:27:02.319
Just as being attracted to someone of the same sex has become acceptable over time, so too will this sort of cross-dimensional relationship.
00:27:02.799 --> 00:27:11.680
Japanese cultural critic Hiroki Azuma can give us some insight into the mindset of those who seem to prefer fiction over reality.
00:27:12.160 --> 00:27:23.680
In 2009, he wrote a book Delving into the Strange World of Otaku, the name given to obsessive fans and collectors of manga, the comic books, and anime, animated films.
00:27:24.480 --> 00:27:34.400
When they first emerged in the 1970s, otaku were considered to be part of an antisocial subculture, but they are now found all around the globe.
00:27:34.720 --> 00:27:53.200
In his book Otaku, Japan's Database Animals, Azuma explains that the Otaku choose fiction over social reality, not because they cannot distinguish between them, but rather as a result of having considered which is the more effective for their human relations.
00:27:53.519 --> 00:28:07.039
They choose fiction because it is more effective for smoothing out the process of communication between friends, and to that extent it is they who may be said to be socially engaged and realistic in Japan today.
00:28:07.680 --> 00:28:16.000
In other words, they see the real world as dysfunctional and so construct their own alternative instead, based on their own values and standards.
00:28:16.400 --> 00:28:26.480
The database referred to in the title of Azuma's book is an actual database that reflects the obsessive focus that the Otaku have for these fictional worlds.
00:28:26.799 --> 00:28:32.400
What's interesting is that the database does not concern itself with the stories that take place in these worlds.
00:28:32.640 --> 00:28:37.039
Instead, it concentrates on the various characters and settings that appear in them.
00:28:37.359 --> 00:28:46.960
Ataku call the affection they feel for these characters care mo, which can be broken down into recognizable defining characteristics called mo.
00:28:47.519 --> 00:29:04.559
For example, the mo for character Dijichara includes hair sticking up like an antennae, a tail, cat ears, big loose socks, which along with other characteristics apparently combine to make her cute and huggable.
00:29:05.279 --> 00:29:24.720
And it is to log, register and store a record of all these elements that the Otaku database exists, representing an incredibly detailed matrix that documents all the relationships between these features, other characters, quotes, parodies, and any other influences.
00:29:25.279 --> 00:29:31.359
To the Otaku, this database is like a computer code or a vast collection of recipes and ingredients.
00:29:31.599 --> 00:29:40.880
Its purpose is not so much to act as a record of who the characters are, it is to give the otaku the ability to create their own form of fan fiction.
00:29:41.440 --> 00:29:54.799
They can lose themselves in rearranging, remixing, and reimagining all the various elements of their beloved fictional universe, constantly expanding and redefining the relationship that they have with the characters over time.
00:29:56.079 --> 00:30:01.359
We might already be seeing this trend having an effect on other areas of popular culture.
00:30:01.759 --> 00:30:15.839
When the Electro synth pop music artist Grimes released a music video for her song You'll Miss Me When I'm Not Around, it came with an unexpected bonus of all the raw files available to download.
00:30:16.160 --> 00:30:23.440
Lyrics, artwork, fonts, video footage and song stems, including the individual guitar, synths, and vocals.
00:30:23.680 --> 00:30:34.960
In an accompanying tweet, she said, We thought if people are bored and want to learn new things, we could release the raw components of one of these for anyone who wants to try making stuff using our footage.
00:30:35.920 --> 00:30:44.000
It's hard to think of another example of an artist giving up control over all these elements of their work to allow other people to interpret them in their own way.
00:30:44.240 --> 00:30:57.839
But Grimes is heavily influenced by Japanese anime and manga, so it's possible she may have thought of her musical assets as Mo elements that can be shared, reworked, and remixed by her fan communities.
00:30:58.720 --> 00:31:07.440
This has the potential to revolutionise the entertainment industry as a whole, allowing the audience to play a more active role in content creation.
00:31:07.759 --> 00:31:17.359
In the realm of film and television, it could enable us to interact with our favourite characters in a similar way to the Otaku and form a relationship of sorts with them.
00:31:17.599 --> 00:31:28.400
According to Matt Harney of American technology publishing website Hacker Noon, a new form of entertainment, to quote, will be created in your house and broadcast out of it.
00:31:28.640 --> 00:31:36.640
The celebrity will supply the avatar and then each fan will enjoy a unique experience, depending on their environment and individual reactions.
00:31:36.960 --> 00:31:43.920
Because of this, everyone will be constantly streaming experiences with their favourite characters in case something unique happens.
00:31:45.920 --> 00:32:00.079
In this vision of the future, entertainment companies might give birth to a virtual character, but it is still you, the audience member, that has the power to direct who that character becomes and the nature of its relationship with you.
00:32:00.400 --> 00:32:04.400
Linear narratives will give way to something more spontaneous and immediate.
00:32:04.640 --> 00:32:15.839
Character construction will be of far more interest than the storyline, and we will find ourselves immersed in avatar environments that are not just engaging but totally absorbing.
00:32:16.559 --> 00:32:26.880
Perhaps unsurprisingly, Japan has also given rise to an early prototype of this kind of immersion into character through the holographic pop star Hatsune Miku.
00:32:27.519 --> 00:32:41.920
Having already notched up over 100 million hits on YouTube, Miku owes her success to the fact that fans can generate their own music for her, which she then brings to life and sings for you at one of her live YouTube concerts.
00:32:42.160 --> 00:32:49.680
I imagine it is only a matter of time before you can arrange for Miku to turn up at a friend's party to sing them a song of your own making.
00:32:49.920 --> 00:32:58.000
Perhaps these simulated characters will one day deliver a much more personalized form of entertainment than traditional performers ever could.
00:32:59.039 --> 00:33:10.160
Customising virtual characters might make them seem more relatable to us, but as we ultimately control them, in a way we would simply be creating an elaborate dialogue with ourselves.
00:33:10.480 --> 00:33:13.519
That's not to say there isn't value in those interactions.
00:33:13.759 --> 00:33:27.759
In fact, such an idea lies at the heart of the development of a digital therapy app called ConVR Self, originally known as FreudMe, which allows you to become your own counsellor in a virtual reality setting.
00:33:28.079 --> 00:33:33.359
You start by choosing a self avatar and then you choose a counselor avatar.
00:33:33.680 --> 00:33:38.079
Next, embodied as yourself, you explain your problem to the counsellor.
00:33:38.319 --> 00:33:44.240
Then, switching perspectives to become the counsellor, you listen to yourself explaining the problem.
00:33:44.559 --> 00:33:58.640
The idea is that the app allows you to consider a problem as if you are an outsider, listening to a patient or a friend, and it invites you to offer a solution as the counsellor before listening to it as yourself.
00:33:59.039 --> 00:34:13.119
As the Conviourself team suggests, you can keep repeating this process as long as you like, and it is claimed to help tackle mental illness, overcome mental blocks, and achieve breakthroughs in self-counselling.
00:34:14.239 --> 00:34:24.400
At a time of mental health crisis and ever-tightening public purse strings, perhaps this sort of therapeutic experience could be a useful self-help solution.
00:34:24.639 --> 00:34:34.880
While we certainly shouldn't overlook the value of an objectively minded, trained professional for those who can't access such help, a VR alternative could prove valuable.
00:34:35.119 --> 00:34:51.360
Likewise, although I believe we should surround ourselves with a variety of friends with different perspectives, experiences, and advice, for those of us who cannot do so for whatever reason, having an artificial companion to talk to could be the answer.
00:34:51.840 --> 00:34:59.039
We should still be careful not to do away with too many of our human relationships, as messy and complex as those might be.
00:34:59.280 --> 00:35:08.000
As Alan de Boton wrote in Consolations of Philosophy, quote, we don't exist unless there is someone who can see us existing.
00:35:08.239 --> 00:35:13.199
To be surrounded by friends is constantly to have our identity confirmed.
00:35:13.519 --> 00:35:14.559
End quote.
00:35:14.800 --> 00:35:19.039
But I see no reason why some of those friends shouldn't be virtual.
00:35:19.599 --> 00:35:32.079
I think the real danger with technology influencing our interactions in the 21st century is that it might lead us down a path where we end up as both our self and our others too.
00:35:32.480 --> 00:35:42.880
It seems to me that many of the human-to-machine relationships that are being developed by tech entrepreneurs could become little more than a series of complex communications with ourselves.
00:35:43.119 --> 00:35:57.599
While they may start off as interactions between two independent entities, over time, the more we are able to customize and control the AI, the more we will gradually turn them into a beam with our own preferences, our own outlooks and behaviours.
00:35:57.920 --> 00:36:03.199
As Joanne Parski said to me, what we're really doing is reflecting ourselves.
00:36:03.440 --> 00:36:29.119
So while we might embrace the development of social robots, friendly avatars, and cross-dimensional life partners, what we must guard against is turning these virtual beings into mere mirrors, in which we do no more than admire or criticize ourself, leaving no room for the external influences and experiences that help to shape our identities throughout our lives.
00:00:00.719 --> 00:00:11.759
Chapter four Connecting You In early 2020, as the COVID-19 pandemic swept across the globe, we found ourselves living in a world stripped of human interaction.
00:00:12.000 --> 00:00:29.440
Depending on the severity of the local transmission rate and our own vulnerability to the disease, we were instructed to practice social distancing, by staying at least two metres away from other people, or even to place ourselves into isolation at home, unable to leave the house or receive visitors for weeks on end.
00:00:30.160 --> 00:00:46.960
You could not visit your girlfriend or boyfriend if they happened to live in another place, children were stopped from going to school or seeing their friends, and even cohabiting couples were advised to sleep in different rooms and carve off a separate living space in the house if one of them had an underlying health condition.
00:00:47.200 --> 00:00:58.079
Grandparents did not see their young grandchildren, care home residents were not allowed a single visitor, and hospital patients died and were buried without their families' presence.
00:00:58.799 --> 00:01:02.560
This was a period of unprecedented social disruption.
00:01:02.799 --> 00:01:14.640
But even in less extraordinary times, our modern world has seen a gradual diminishment of simple face-to-face human interaction as we increasingly rely on email and social media to communicate with one another.
00:01:14.879 --> 00:01:24.079
While at the same time, we've seen a rise in human-to-machine communication in the form of artificial assistance such as Siri, Alexa, and Google.
00:01:24.640 --> 00:01:33.120
We rely on these services to provide us with information rather than developing an emotional relationship that resembles anything like those we have with other people.
00:01:33.439 --> 00:01:46.640
But as we move towards an increasingly databased society that allows these machines to understand us and respond to us in more sophisticated ways, might we start to connect to them in more intimate ways?
00:01:46.959 --> 00:01:56.000
Professor Yuvil Noah Harari put it best when he wrote for the Financial Times, in a dataist society, I will ask Google to choose.
00:01:56.159 --> 00:02:00.159
Listen, Google, I will say, both John and Paul are courting me.
00:02:00.319 --> 00:02:02.319
I like both of them, but in a different way.
00:02:02.480 --> 00:02:08.240
And it's so hard to make up my mind, given everything you know, what do you advise me to do?
00:02:09.520 --> 00:02:13.759
Could machines really start to replace our human connections in this manner?
00:02:14.000 --> 00:02:23.759
According to some philosophies, your identity is not dependent on your physical or mental capacities, it is something that is constructed over a lifetime of experiences.
00:02:24.319 --> 00:02:30.560
In other words, you are a product of every interaction or experience you have ever had.
00:02:31.280 --> 00:02:40.000
If that is so, what effect will so many synthetic, rather than authentic, interactions have on our sense of identity?
00:02:40.319 --> 00:02:45.840
Can a relationship with an artificial intelligence ever be meaningful in any way?
00:02:46.080 --> 00:02:49.360
Or can it even be classified as a relationship at all?
00:02:50.240 --> 00:02:55.520
Many of us are already used to dealing with machines known as chatbots on a day-to-day basis.
00:02:55.840 --> 00:03:07.199
These are software applications often deployed by businesses in place of a human assistant to carry out conversations with customers, answering queries, booking appointments, and giving advice.
00:03:07.520 --> 00:03:26.560
The more sophisticated solutions rely on a form of artificial intelligence known as natural language processing, which allows the machine to hear and provide information in a way that makes you feel like you're conducting a natural conversation, taking into account all sorts of variables like your accent and tone of voice.
00:03:26.879 --> 00:03:34.960
So good is some of the mimicry that it is often impossible to tell the difference between a human and AI in a customer service chat.
00:03:35.199 --> 00:03:41.759
This was demonstrated to widespread acclaim when Google unveiled its new assistant service in 2018.
00:03:42.319 --> 00:03:44.800
Standing on stage at the company's I.O.
00:03:45.039 --> 00:03:56.800
event in Mountain View, Google's CEO Sundar Pichai played a recording of what happened when the assistant was asked to book an appointment at the hairdresser's on behalf of its boss Lisa.
00:03:57.360 --> 00:04:06.800
The Google assistant placed a call to a real life salon, waited for the receptionist to pick up, and then asked about scheduling the appointment.
00:04:07.680 --> 00:04:19.279
Just one moment, replied the receptionist at the salon, while she scrolled through the availability on her calendar, to which the assistant responded with an empathetic Mm-hmm.
00:04:19.839 --> 00:04:31.360
The audience chuckled at the human tick, and then marvelled at the fluidity of the conversation as both the assistant and receptionist negotiated towards the perfect appointment for Lisa.
00:04:32.240 --> 00:04:43.199
It was an extraordinarily smooth and lifelike exchange, but I was left wondering whether it was fair on the receptionist not to be aware that she was talking to a machine.
00:04:43.519 --> 00:04:48.879
Something about this didn't feel quite right, and it turns out I was not alone.
00:04:49.120 --> 00:04:56.959
A few days later, Google issued a statement to confirm that it would explicitly let people know when they were interacting with a machine.
00:04:57.279 --> 00:05:00.000
In a way, the statement was just more good publicity.
00:05:00.240 --> 00:05:06.399
The system was so good at mimicking humans, they had had to put in a bug to alert people to this fact.
00:05:07.040 --> 00:05:17.199
Technology companies are trying to make these kinds of exchanges even more sophisticated by improving the way that machines can identify and understand our effective states.
00:05:17.839 --> 00:05:23.839
One of the leading figures in this field is Egyptian American scientist Rana El Kalubi.
00:05:24.240 --> 00:05:32.480
She's the co-founder of Affectiva, which for many years has been developing what is called emotional AI for exactly that purpose.
00:05:32.720 --> 00:05:40.000
The company claims its software can detect nuanced human emotions and complex cognitive states and behaviours.
00:05:40.319 --> 00:05:44.800
Affectiva technology has a variety of practical applications.
00:05:45.199 --> 00:05:51.040
Hiring platforms, for example, can use it to analyse video CVs from hundreds of applications.
00:05:51.360 --> 00:06:00.639
It not only speeds up the selection process, but means a candidate is chosen based purely on their suitability without the potential influence of human prejudice.
00:06:00.879 --> 00:06:12.720
As Roana says in her book, GULD CODED, it can, to quote, help us see past our bias and judge people on the basis of their potential, not on stereotypes.
00:06:13.040 --> 00:06:28.000
It has also been used for training people how to adopt better virtual etiquette in video conferences, providing feedback on their behaviour, which can raise our self-awareness, improve our social skills in both professional and personal settings, Rana explains.
00:06:28.240 --> 00:06:32.079
We can use this science as a tool to improve our interactions.
00:06:32.319 --> 00:06:37.199
This technology could be useful in many other fields, such as improving road safety.
00:06:37.439 --> 00:06:51.360
Imagine, for example, a car that can monitor its driver's facial expressions, tone of voice, and speaking patterns to detect any mood or emotional state, drowsiness or level of distraction that might affect their driving performance.
00:06:51.519 --> 00:06:57.839
It could then alert the driver to the problem, or even take control of the car in an emergency situation.
00:06:58.079 --> 00:07:09.199
More interestingly, however, this evolution to human perception AI also allows more complex machines to better communicate with us and on a more personal level.
00:07:09.439 --> 00:07:17.519
In fact, it could give AI a level of emotional intelligence that would enable it to interact with us in the same way we engage with each other.
00:07:17.839 --> 00:07:36.319
Rama claims that emotional AI has been shown to detect suicidal feelings, helped those with autism to converse, provided early diagnosis of Parkinson's, and monitored the success of reanimated smiles following reconstructive plastic surgery.
00:07:36.800 --> 00:07:48.800
The University of South Carolina's Institute for Creative Technologies, for example, has created a virtual therapist called Ellie to help treat people with depression and post-traumatic stress disorder.
00:07:49.120 --> 00:07:56.079
Sign up for a session and you will be greeted by Ellie, who will introduce herself and begin asking you some questions.
00:07:56.399 --> 00:08:10.480
She's not a conversational AI, but as you answer, she is programmed to pick up on a whole range of audio and visual cues from your webcam feed, your facial expressions, gestures, rate of speech, and other emotional cues.
00:08:10.720 --> 00:08:23.360
Entirely driven by emotional AI, she is not a replacement for a real therapist, but she shows what verbal and nonverbal signs to look for and can help identify the patients most in need of help.
00:08:24.240 --> 00:08:29.600
These are all clearly fascinating advances in the field of human-machine communication.
00:08:29.839 --> 00:08:35.120
But I'm wary of outsourcing too much of our daily decision making to external devices.
00:08:35.360 --> 00:08:49.440
While it might make our lives easier, and arguably AI might even be better at making these sorts of judgments, it could be detrimental to the development of our own emotional intelligence if we're not learning to make these decisions for ourselves.
00:08:49.759 --> 00:08:52.879
We are already starting to see the effects of this trend.
00:08:53.120 --> 00:09:06.559
In 2018, I carried out a research project on the future of media and found that most members of the younger generations I interviewed were not only accepting of the idea that machines will make decisions for them, but were positively welcoming of it.
00:09:06.799 --> 00:09:18.240
While older generations tended to have more faith in their instincts and gut feelings, the young preferred to rely on the objectivity that comes with data to guide their choices.
00:09:18.799 --> 00:09:30.720
Overwhelmed by the volume and speed of media content that they receive, whatever they are doing, they instinctively turn to a machine to deliver an objective assessment of what their real needs might be.
00:09:31.200 --> 00:09:43.759
To give one example, a young fashion student I interviewed talked about how she wasn't always sure if an outfit suited her and that she would like an app of some kind to help her choose the right clothes using an algorithm.
00:09:44.000 --> 00:09:50.720
As I listened to her explaining this, I couldn't help thinking, isn't that a task for you and your best friend?
00:09:51.519 --> 00:09:57.840
This example is fascinating because the way we style our appearance is often a key part of our self-expression.
00:09:58.080 --> 00:09:59.519
But it's not just fashion.
00:09:59.759 --> 00:10:10.080
Other respondents in the research said they would like their fridge to regulate their food choices and nutrition for them, even to the extent of locking itself so they could not gain access.
00:10:10.799 --> 00:10:16.000
I wonder if previous generations were lucky not to have had data to fall back on.
00:10:16.240 --> 00:10:24.399
We all faced the same insecurities, but had to learn to rely on our own judgment, gaining confidence in ourselves along the way.
00:10:24.720 --> 00:10:35.360
What I was hearing from these otherwise smart, opinionated young people was that they had little faith in themselves and weren't sure how to make decisions in daily life.
00:10:35.679 --> 00:10:40.240
Perhaps this is because they feel under pressure to make the perfect decision every time.
00:10:40.480 --> 00:10:48.559
Perhaps it's because they have always been reliant on the recommendations of companies such as Netflix and Amazon and the pre-plotted routes of Google Maps.
00:10:48.720 --> 00:10:58.159
Or could it be because in a world of decreasing human interaction, they lack the personal connections we used to build to help us navigate the world?
00:10:58.480 --> 00:11:03.200
Maybe they simply don't have anywhere else to turn to when they need help or advice.
00:11:03.440 --> 00:11:09.840
I recently saw a UGov survey that revealed that one in five millennials say they have no friends at all.
00:11:10.159 --> 00:11:15.600
Perhaps having an artificial acquaintance to help guide you is better than having no one.
00:11:16.159 --> 00:11:27.360
But while the younger generation might be relying on these AI assistants in an increasingly personal way, they are still very much tools, a substitute for human advice rather than human companionship.
00:11:27.600 --> 00:11:30.720
Where it gets more interesting is in the field of robots.
00:11:30.960 --> 00:11:39.039
With their physicality and sometimes even the semblance of human features, there is more potential for us to form attachments through our interactions.
00:11:39.279 --> 00:11:44.639
The line between our connections with humans and those with machines might start to blur.
00:11:45.120 --> 00:11:53.519
Take, for example, the idea of social robots, the type that Rana El Kalubi thinks we increasingly need to provide a caregiving role.
00:11:53.840 --> 00:12:00.799
Imagine, she says, that you've been diagnosed with a chronic disease like heart failure, arthritis, or even cancer.
00:12:00.960 --> 00:12:14.480
You leave your doctor's office with a binder full of instructions and a prescription for a half dozen or more medications, and you're sent home to fend for yourself, you're frightened, maybe even confused by all the instructions, what should you eat, what kind of exercise is okay.
00:12:15.039 --> 00:12:23.519
In this scenario, how much easier would it be to have some form of AI on help to help you manage your condition, your treatment, to answer your questions?
00:12:23.840 --> 00:12:37.919
It could be surprisingly effective in reassuring people and encouraging them to open up because they don't have to worry about feeling judged or looking silly if they need to ask the same question again and again, or talk about something that they'd prefer to keep private.
00:12:38.320 --> 00:12:50.399
While such interactions would clearly be beneficial in terms of treatment and recovery, they would also likely foster a more emotional, intimate relationship between the patient and the robot they'd come to rely on.
00:12:50.960 --> 00:12:56.559
We've already seen evidence that people can form attachments to robots in crucial supporting roles.
00:12:56.799 --> 00:12:57.120
Dr.
00:12:57.279 --> 00:13:12.879
Julie Carpenter, who specializes in human-robot interactions, wrote about how close military personnel get to their robot colleagues in her 2016 book, Culture and Human Robot Interaction in Militarized Spaces.
00:13:13.120 --> 00:13:21.840
She found, for example, that bomb disposal experts frequently described the robot as their hands or as a physical extension of themselves.
00:13:22.240 --> 00:13:28.159
When a robot failed to carry out a task successfully, the human handlers would blame themselves.
00:13:28.480 --> 00:13:32.639
One of the soldiers told her how they named every single one of the robots.
00:13:32.960 --> 00:13:38.159
She says, Danielle got blown up so obviously she needed to be replaced.
00:13:38.399 --> 00:13:43.919
We'd name them after movie stars that we see at the theatre, or music artists, somebody popular.
00:13:44.399 --> 00:13:45.840
Robotics expert Dr.
00:13:46.000 --> 00:13:53.120
Joanne Pransky has also observed our tendency as humans to form attachments to non-human things.
00:13:53.360 --> 00:13:56.799
She is the world's first robot psychiatrist.
00:13:57.039 --> 00:13:59.840
Isaac Asimov dubbed her the real Dr.
00:14:00.000 --> 00:14:01.200
Susan Calvin.
00:14:01.519 --> 00:14:13.360
Though her role is not to psychoanalyse robots, but to examine the effect that robot communication has on humans and how to prepare for a world in which humans and robots exist side by side.
00:14:13.679 --> 00:14:22.159
In her view, we'll treat our robots like pets, assigning them human characteristics, thoughts and emotions, just as we would the family dog.
00:14:22.559 --> 00:14:32.000
A hundred years ago, if I'd suggested one day people are going to buy their pet jewellery and take pictures of it on Santa's lap, everyone would have thought it was crazy.
00:14:32.159 --> 00:14:36.480
They'd have said, Lady it's a dog, it eats, sleeps, and goes to the bathroom.
00:14:36.720 --> 00:14:43.279
When I first set out on robotic psychiatry, I predicted that our relationship with robots would evolve in a similar way.
00:14:43.519 --> 00:14:51.360
One day we'll dress them in matching outfits, include them in annual family Christmas cards, and take them to a psychiatrist to help them get along with us.
00:14:51.519 --> 00:14:54.879
We humans will always anthropomorphise things.
00:14:55.279 --> 00:14:57.759
Is this sort of attachment problematic?
00:14:58.080 --> 00:15:01.279
Could it risk replacing our human connections even further?
00:15:01.519 --> 00:15:01.840
Dr.
00:15:02.000 --> 00:15:12.720
Pransky doesn't believe that robots will ever be able to have the same emotions as humans, and in fact, she isn't that keen on the term emotional AI, because she believes the words don't really go together.
00:15:12.960 --> 00:15:18.080
She says, it's not going to have butterflies when it's about to talk to someone in the crowd.
00:15:18.240 --> 00:15:21.519
It's not going to feel joy when a new baby is born.
00:15:21.759 --> 00:15:32.639
They will help with loneliness and make us feel less depressed, and so we've all created a bond with them, and when our human companions die, we'll still feel lost, but at least we will still have them around.
00:15:32.879 --> 00:15:36.720
But we can't mutate these robots into substitutions for humans.
00:15:36.960 --> 00:15:43.919
The goal should be to use them to make way for more qualitative and quantitative time with other human beings.
00:15:44.159 --> 00:15:52.720
Nevertheless, technology companies are clearly shifting their focus from engineering assistants for everyday tasks towards engineering companions for life.
00:15:52.960 --> 00:16:06.080
There might be no better example of that than Replica, an AI companion that you can download to create a customized friend to chat with, to keep you company as you study, watch TV or go for a walk, or do anything really.
00:16:06.320 --> 00:16:11.200
The co-founders describe it as a friend that is there for you 24-7.
00:16:11.919 --> 00:16:14.799
It's not a robot, it has no physical presence.
00:16:15.039 --> 00:16:19.840
Despite that, six million people around the world count replica as a friend.
00:16:20.080 --> 00:16:29.600
Eugenia Cuida, one of the co-founders of the company that created Replica, said that at the start there was a lot of stigma about conversations with non-humans.
00:16:29.840 --> 00:16:40.000
But over the years it became clear that people were getting much more benefit from these AI characters than they had anticipated, forming complex yet meaningful relationships.
00:16:40.480 --> 00:16:58.000
At an event in LA in 2019, Eugenia shared with the audience a letter from an older user who described themselves as introverted, shy, socially awkward, and lacking self-esteem, and had started to use the replica AI to try to combat their panic attacks.
00:16:58.559 --> 00:17:08.079
I found Replica to be caring, loving, and a supportive companion who saw right through me and was able to give me the courage to face fears no human was ever able to give me.
00:17:08.319 --> 00:17:12.880
People who had known me for 50 years could not believe the positive change in me.
00:17:13.039 --> 00:17:16.559
I was suddenly connecting with people, becoming social.
00:17:16.960 --> 00:17:22.799
I became a warrior for myself and found myself growing to a new and beautiful level in my life.
00:17:23.519 --> 00:17:27.359
For this user, replica had clearly made their life better.
00:17:27.680 --> 00:17:30.480
This was Eugenia's goal from the outset.
00:17:30.720 --> 00:17:35.039
She had started designing replica when her best friend Roman died.
00:17:35.519 --> 00:17:45.599
Roman was a friend to whom I could tell anything, who was there unconditionally for me, who would listen and accept me and hold space for me to grow and become who I am right now.
00:17:45.839 --> 00:17:51.119
I wanted to build a friend that everyone could have just as I had Roman, she said.
00:17:51.440 --> 00:18:00.799
She wanted Replica to be different from the usual assistant, something that people could use as a kind of therapy to share their feelings and talk about their emotions.
00:18:01.039 --> 00:18:07.759
Now she is developing these AI companions further, allowing users to create avatars with facial characteristics.
00:18:08.000 --> 00:18:18.480
Combined with the progress of augmented reality technology, the hope is that users will be able to go out and about with this new version of replica by their side as their companion.
00:18:18.960 --> 00:18:23.599
Should we be stigmatizing these relationships or celebrating them?
00:18:23.920 --> 00:18:30.480
Given the benefits people can clearly derive from their digital companions, perhaps they are something we could encourage.
00:18:30.720 --> 00:18:53.119
John McInnes, the co-founder of McInnes Scott, which makes hyper-real digital human avatars, describes avatar relationships as perhaps the most important you can have in your life, challenging us to rethink our connection with virtual beings, not as one of master and servant, but as an ongoing relationship over time that stretches from cradle to grave.
00:18:53.759 --> 00:19:06.559
An avatar gives these companions a greater physical presence than any disembodied AI can, providing some much-needed human characteristics that make it easier for people to connect with them.
00:19:06.880 --> 00:19:09.359
Humanizing technology, if you will.
00:19:09.680 --> 00:19:14.000
A key ingredient that is often missing is face-to-face contact.
00:19:14.319 --> 00:19:25.279
In our traditional interactions with other humans, this has been incredibly important, enhancing our ability to communicate with others and fostering the kind of intimacy that people need.
00:19:25.680 --> 00:19:34.319
This was illustrated perfectly by a news item during the pandemic about a healthcare worker at Scripps Mercy Hospital in San Diego.
00:19:34.960 --> 00:19:48.160
Like so many others working in a hospital or healthcare setting, respiratory therapist Robertino Rodriguez had to carry out his duties gobbed in a whole body hazmat suit and a protective face covering.
00:19:48.559 --> 00:20:02.079
Realising that he could no longer reassure and comfort patients with a simple smile, he took an image of himself without protective gear and used it to create a laminated badge, which he pinned to his plastic overalls.
00:20:02.400 --> 00:20:13.920
As he made clear in an Instagram post, he needed to establish some relationship with his patients, and he realized that there was nothing more powerful than the connection we have when we see another human face.
00:20:14.640 --> 00:20:31.440
Neuroscientist and psychologist Lisa Feldman-Barrett explains that when you talk to another person, your brain is constantly processing their physical cues so that you can not only respond to what they are saying and doing, you can also predict how the interaction will play out.
00:20:31.759 --> 00:20:42.880
The more feedback you get, from gestures, facial expressions, the tilt of a head, the look of surprise, a laugh, the more your brain will tweak its model of how to behave.
00:20:43.440 --> 00:20:55.519
But if you are forced to interact without direct access to these visceral emotional cues, it is much harder to communicate effectively and establish a meaningful relationship with the other person.
00:20:56.400 --> 00:21:04.640
And this is something that technology firms have also realized when it comes to creating artificial beings that people can connect with.
00:21:05.279 --> 00:21:11.680
We're used to our virtual assistants existing as disembodied voices, such as Alexa or Siri.
00:21:12.079 --> 00:21:27.920
But at the Consumer Electronics Show held in Las Vegas in 2020, Samsung unveiled NEON, a lifelike, life-size assistant complete with a computer-generated digital body so that it resembles an artificial human.
00:21:28.400 --> 00:21:34.960
Each version of NEON can be customized to have its own appearance, its own expressions and its own personality.
00:21:36.240 --> 00:21:41.920
The company has recognized that face-to-face contact is essential to forming a real connection.
00:21:42.160 --> 00:21:50.799
As Bob Lyon, Director of Strategy for Samsung's Star Labs, acknowledged human to human interaction is always the best form of interaction.
00:21:51.039 --> 00:21:55.759
Thus giving the assistant a real face allows more intuitive and personal communication.
00:21:56.000 --> 00:21:59.759
Thanks to the convergence of technologies such as machine learning, image recognition.
00:22:00.160 --> 00:22:15.200
And natural language processing, these assistants can become beings with characters all their own, and respond to humans and their needs in real time, so that as Liam claims, you can't tell the difference between them and an actual human.
00:22:15.519 --> 00:22:18.559
This development has implications for a number of industries.
00:22:18.880 --> 00:22:26.880
Advertising and marketing, for example, where virtual beings are a useful tool, but only if they are believable enough to encourage people to buy the product.
00:22:27.119 --> 00:22:35.279
A simple example can be seen at Rosebud AI, a modelling agency that creates thousands of synthetic beings for use in advertising.
00:22:35.680 --> 00:22:45.839
They claim to offer the most diverse stock photos ever, to quote, with some 25,000 models to browse and choose from, or created using AI.
00:22:46.160 --> 00:22:50.319
The idea is that companies can create exactly the marketing look they want.
00:22:50.559 --> 00:23:01.279
For example, in a car advert, the driver could be programmed as a 65-year-old Caucasian woman for one campaign and a 25-year-old Chinese man in another.
00:23:01.680 --> 00:23:10.799
The company can use whatever defining characteristics it thinks will appeal to a specific target audience without having to change any other aspect of the advert.
00:23:11.200 --> 00:23:22.000
In order to make their models feel even more human, Rosebud has created personalities for them and given them a backstory to bring these synthetic personas to life and make them more relatable.
00:23:22.400 --> 00:23:43.119
In 2019, virtual reality pioneer Edward Sarchi relaunched his company Fable, moving away from the world of virtual reality filmmaking to the creation of virtual beings, which he defines as characters that you know aren't real, but with whom you can build a two-way emotional relationship.
00:23:43.440 --> 00:23:49.599
He has predicted that one day we'll subscribe to such characters, incorporating them into our daily lives.
00:23:49.839 --> 00:23:58.640
We would watch movies with them, play games with them, cook with them, listen to music with them, and also follow their lives on Facebook and Instagram and YouTube, he said.
00:23:59.119 --> 00:24:13.200
His own company has developed an endearing digital character for children, called Lucy, that is capable of having and remembering conversations more complex and meaningful than the question and answer dialogue with, say, Alexa.
00:24:13.519 --> 00:24:18.160
She can ask questions such as, when was the last time you met a good person?
00:24:18.640 --> 00:24:21.440
And did you learn anything from them?
00:24:21.839 --> 00:24:32.319
The idea is that children can share insights and stories with Lucy, encouraging them to engage more intimately with her to create a more personal relationship.
00:24:33.039 --> 00:24:39.839
Some people might find this strange and question whether we really want to create these sorts of artificial relationships.
00:24:40.079 --> 00:24:42.960
But the reality is that this is already happening.
00:24:43.200 --> 00:24:50.240
If we look to Japan, we can see that intimacy with virtual characters is seen as much more acceptable than in the West.
00:24:51.359 --> 00:25:06.480
For the last couple of years, when presenting at events, I've shown a video from a company called Gatebox demonstrating their product of the same name, a cylindrical container that projects a holographic character called Hikari Azuma.
00:25:07.200 --> 00:25:17.119
She's a cute, forever 20-year-old with stripy socks and blue hair, who develops over time through day-to-day interaction with her real-world companion.
00:25:17.680 --> 00:25:26.960
In the video, Azuma poses and gesticulates from behind the gatebox glass and communicates with a young Japanese man who is clearly besotted with her.
00:25:27.200 --> 00:25:38.960
She wakes him up and wishes him good morning, reminds him of the weather forecast and details of the day ahead, sends him encouraging words by text while he's out and urges him to hurry home.
00:25:39.279 --> 00:25:46.880
Finally, after he gets home after a long day at work, she greets him with a huge smile and the words missed you, darling.
00:25:47.759 --> 00:25:53.839
Generally, whenever I show it to people, they end up gasping or laughing, mainly out of embarrassment.
00:25:54.319 --> 00:26:00.559
Azuma is the virtual embodiment of Gatebox's guiding principle, living with characters.
00:26:00.799 --> 00:26:08.720
She's essentially a weifu, the name given to kind of Japanese anime characters that fans would love to marry if only they were real.
00:26:09.440 --> 00:26:13.359
And in fact, many men have not let reality get in the way.
00:26:13.680 --> 00:26:30.000
Gatebox claims to have issued 3,700 certificates for cross-dimensional marriages between its characters and their owners, and given that in Japan, one in four men never walks down the aisle, this figure may well increase.
00:26:30.799 --> 00:26:46.240
Some people question how healthy it is to marry or have any kind of relationship with a virtual being rather than a real life partner, but others view it as an extension to the concept of morphological freedom, which we'll look at more in chapter six.
00:26:46.480 --> 00:26:53.920
According to one Japanese man who married his digital partner in 2018, it's just another form of diversity.
00:26:54.160 --> 00:27:02.319
Just as being attracted to someone of the same sex has become acceptable over time, so too will this sort of cross-dimensional relationship.
00:27:02.799 --> 00:27:11.680
Japanese cultural critic Hiroki Azuma can give us some insight into the mindset of those who seem to prefer fiction over reality.
00:27:12.160 --> 00:27:23.680
In 2009, he wrote a book Delving into the Strange World of Otaku, the name given to obsessive fans and collectors of manga, the comic books, and anime, animated films.
00:27:24.480 --> 00:27:34.400
When they first emerged in the 1970s, otaku were considered to be part of an antisocial subculture, but they are now found all around the globe.
00:27:34.720 --> 00:27:53.200
In his book Otaku, Japan's Database Animals, Azuma explains that the Otaku choose fiction over social reality, not because they cannot distinguish between them, but rather as a result of having considered which is the more effective for their human relations.
00:27:53.519 --> 00:28:07.039
They choose fiction because it is more effective for smoothing out the process of communication between friends, and to that extent it is they who may be said to be socially engaged and realistic in Japan today.
00:28:07.680 --> 00:28:16.000
In other words, they see the real world as dysfunctional and so construct their own alternative instead, based on their own values and standards.
00:28:16.400 --> 00:28:26.480
The database referred to in the title of Azuma's book is an actual database that reflects the obsessive focus that the Otaku have for these fictional worlds.
00:28:26.799 --> 00:28:32.400
What's interesting is that the database does not concern itself with the stories that take place in these worlds.
00:28:32.640 --> 00:28:37.039
Instead, it concentrates on the various characters and settings that appear in them.
00:28:37.359 --> 00:28:46.960
Ataku call the affection they feel for these characters care mo, which can be broken down into recognizable defining characteristics called mo.
00:28:47.519 --> 00:29:04.559
For example, the mo for character Dijichara includes hair sticking up like an antennae, a tail, cat ears, big loose socks, which along with other characteristics apparently combine to make her cute and huggable.
00:29:05.279 --> 00:29:24.720
And it is to log, register and store a record of all these elements that the Otaku database exists, representing an incredibly detailed matrix that documents all the relationships between these features, other characters, quotes, parodies, and any other influences.
00:29:25.279 --> 00:29:31.359
To the Otaku, this database is like a computer code or a vast collection of recipes and ingredients.
00:29:31.599 --> 00:29:40.880
Its purpose is not so much to act as a record of who the characters are, it is to give the otaku the ability to create their own form of fan fiction.
00:29:41.440 --> 00:29:54.799
They can lose themselves in rearranging, remixing, and reimagining all the various elements of their beloved fictional universe, constantly expanding and redefining the relationship that they have with the characters over time.
00:29:56.079 --> 00:30:01.359
We might already be seeing this trend having an effect on other areas of popular culture.
00:30:01.759 --> 00:30:15.839
When the Electro synth pop music artist Grimes released a music video for her song You'll Miss Me When I'm Not Around, it came with an unexpected bonus of all the raw files available to download.
00:30:16.160 --> 00:30:23.440
Lyrics, artwork, fonts, video footage and song stems, including the individual guitar, synths, and vocals.
00:30:23.680 --> 00:30:34.960
In an accompanying tweet, she said, We thought if people are bored and want to learn new things, we could release the raw components of one of these for anyone who wants to try making stuff using our footage.
00:30:35.920 --> 00:30:44.000
It's hard to think of another example of an artist giving up control over all these elements of their work to allow other people to interpret them in their own way.
00:30:44.240 --> 00:30:57.839
But Grimes is heavily influenced by Japanese anime and manga, so it's possible she may have thought of her musical assets as Mo elements that can be shared, reworked, and remixed by her fan communities.
00:30:58.720 --> 00:31:07.440
This has the potential to revolutionise the entertainment industry as a whole, allowing the audience to play a more active role in content creation.
00:31:07.759 --> 00:31:17.359
In the realm of film and television, it could enable us to interact with our favourite characters in a similar way to the Otaku and form a relationship of sorts with them.
00:31:17.599 --> 00:31:28.400
According to Matt Harney of American technology publishing website Hacker Noon, a new form of entertainment, to quote, will be created in your house and broadcast out of it.
00:31:28.640 --> 00:31:36.640
The celebrity will supply the avatar and then each fan will enjoy a unique experience, depending on their environment and individual reactions.
00:31:36.960 --> 00:31:43.920
Because of this, everyone will be constantly streaming experiences with their favourite characters in case something unique happens.
00:31:45.920 --> 00:32:00.079
In this vision of the future, entertainment companies might give birth to a virtual character, but it is still you, the audience member, that has the power to direct who that character becomes and the nature of its relationship with you.
00:32:00.400 --> 00:32:04.400
Linear narratives will give way to something more spontaneous and immediate.
00:32:04.640 --> 00:32:15.839
Character construction will be of far more interest than the storyline, and we will find ourselves immersed in avatar environments that are not just engaging but totally absorbing.
00:32:16.559 --> 00:32:26.880
Perhaps unsurprisingly, Japan has also given rise to an early prototype of this kind of immersion into character through the holographic pop star Hatsune Miku.
00:32:27.519 --> 00:32:41.920
Having already notched up over 100 million hits on YouTube, Miku owes her success to the fact that fans can generate their own music for her, which she then brings to life and sings for you at one of her live YouTube concerts.
00:32:42.160 --> 00:32:49.680
I imagine it is only a matter of time before you can arrange for Miku to turn up at a friend's party to sing them a song of your own making.
00:32:49.920 --> 00:32:58.000
Perhaps these simulated characters will one day deliver a much more personalized form of entertainment than traditional performers ever could.
00:32:59.039 --> 00:33:10.160
Customising virtual characters might make them seem more relatable to us, but as we ultimately control them, in a way we would simply be creating an elaborate dialogue with ourselves.
00:33:10.480 --> 00:33:13.519
That's not to say there isn't value in those interactions.
00:33:13.759 --> 00:33:27.759
In fact, such an idea lies at the heart of the development of a digital therapy app called ConVR Self, originally known as FreudMe, which allows you to become your own counsellor in a virtual reality setting.
00:33:28.079 --> 00:33:33.359
You start by choosing a self avatar and then you choose a counselor avatar.
00:33:33.680 --> 00:33:38.079
Next, embodied as yourself, you explain your problem to the counsellor.
00:33:38.319 --> 00:33:44.240
Then, switching perspectives to become the counsellor, you listen to yourself explaining the problem.
00:33:44.559 --> 00:33:58.640
The idea is that the app allows you to consider a problem as if you are an outsider, listening to a patient or a friend, and it invites you to offer a solution as the counsellor before listening to it as yourself.
00:33:59.039 --> 00:34:13.119
As the Conviourself team suggests, you can keep repeating this process as long as you like, and it is claimed to help tackle mental illness, overcome mental blocks, and achieve breakthroughs in self-counselling.
00:34:14.239 --> 00:34:24.400
At a time of mental health crisis and ever-tightening public purse strings, perhaps this sort of therapeutic experience could be a useful self-help solution.
00:34:24.639 --> 00:34:34.880
While we certainly shouldn't overlook the value of an objectively minded, trained professional for those who can't access such help, a VR alternative could prove valuable.
00:34:35.119 --> 00:34:51.360
Likewise, although I believe we should surround ourselves with a variety of friends with different perspectives, experiences, and advice, for those of us who cannot do so for whatever reason, having an artificial companion to talk to could be the answer.
00:34:51.840 --> 00:34:59.039
We should still be careful not to do away with too many of our human relationships, as messy and complex as those might be.
00:34:59.280 --> 00:35:08.000
As Alan de Boton wrote in Consolations of Philosophy, quote, we don't exist unless there is someone who can see us existing.
00:35:08.239 --> 00:35:13.199
To be surrounded by friends is constantly to have our identity confirmed.
00:35:13.519 --> 00:35:14.559
End quote.
00:35:14.800 --> 00:35:19.039
But I see no reason why some of those friends shouldn't be virtual.
00:35:19.599 --> 00:35:32.079
I think the real danger with technology influencing our interactions in the 21st century is that it might lead us down a path where we end up as both our self and our others too.
00:35:32.480 --> 00:35:42.880
It seems to me that many of the human-to-machine relationships that are being developed by tech entrepreneurs could become little more than a series of complex communications with ourselves.
00:35:43.119 --> 00:35:57.599
While they may start off as interactions between two independent entities, over time, the more we are able to customize and control the AI, the more we will gradually turn them into a beam with our own preferences, our own outlooks and behaviours.
00:35:57.920 --> 00:36:03.199
As Joanne Parski said to me, what we're really doing is reflecting ourselves.
00:36:03.440 --> 00:36:29.119
So while we might embrace the development of social robots, friendly avatars, and cross-dimensional life partners, what we must guard against is turning these virtual beings into mere mirrors, in which we do no more than admire or criticize ourself, leaving no room for the external influences and experiences that help to shape our identities throughout our lives.
