THREE.
Calculations. Calculations beyond the count, even beyond the imagination of the average person, were being crunched by supercomputers. There were several of them, each one dedicated to a different task that, when working together, amounted to the extraordinary feat of emulating a little girl and an environment for her to grow up in. The supercomputer that rendered her world was currently calculating (among too many other things to list here) the effects of wind and weakening bonds between twig and leaves. Zeros and ones in seemingly infinite long strings ran through its memory, calculating for wind turbulence and other physics models.
These supercomputers were diligently monitored by teams of scientists. Some were specialists in computers, others in cognitive sciences, still others in child psychiatry. Many different fields of expertise, but all united in a common vision. Finally, computing power and knowledge of the design and functions of brains had reached a level of complexity where the prospect of building and simulating a biologically-accurate virtual human was no longer science fiction.
Dr Giulio Dinova, who headed the research team, sat in what had become known as the ‘Sensorium’. That was the name given to the room where sophisticated virtual reality tools translated the abstract mathematical models into something a person could more easily understand. Here, you could enter the world of Emily. You could be a fly-on-the-wall, observing her as she interacted with her world. You could zoom in, right down to the molecular level, and track the neurological pathways that developed as she learned a new task.
Dr Dinova contemplated the past developments that had lead to this little girl living her virtual life. Past researchers had developed sophisticated network models of the metabolic, immune, nervous and circulatory systems. Others had designed structural models of the heart and other muscles. Perhaps most importantly, a team lead by someone called Henry Markram had shown that you could model a brain in software. The limitations of computer power back then had meant only a rat brain could be fully modelled. But (as Markram had suspected it would) reverse-engineering the rat neocortex had given computer scientists new insights into how the power and performance of computers might be dramatically extended.
But, even now with the advent of handheld devices capable of petaflop levels of number crunching, it was still too expensive and time-consuming to build a VR human from scratch. The average person possessed a VR twin constructed from a library of averaged mathematical models of newborns. When a real person was born, non-invasive medical scanning techniques recorded their vital statistics, and that data was then integrated with a model based on such things as sex, ethnicity, geographic origin, and other salient features. The result was a model that closely resembled the actual person. Whenever that person went for medical checkups, his or her virtual twin would be updated with the latest biomedical information. Whenever a person fell ill, the condition would be replicated in the VR twin, and simulations for the range of available treatments would be undertaken, in order to anticipate short and long-term effects. The availability of VR twins had not only eliminated the need for animal experiments, it had also largely consigned the prospect of side-effects from medical treatment to the dustbin of history.
That a VR twin might be suffering in the name of science was not a thought that crossed anyone’s mind. This was because the model was not sufficiently complex to enable an empathetic and aware mind to emerge. The VR twin was nothing more than the integration of past subsystem models into one systematic model which was sufficient for modelling the effects of drugs, but not sufficient to model the subjectivity of pain and suffering, or any qualia for that matter. Simply put, a VR twin was a zombie.
Emily, though, was very different. She was the result of the most sophisticated biophysical and neurological modelling of a person that had ever been attempted. Dr Dinova and his research team had at their disposal supercomputers of seemingly limitless power, not to mention decades-worth of data that provided exhaustive details of the reverse-engineered brain, nervous system, and other parts of a human body. Reading through the blog posts and watching posted video comments, the team (who never ventured outside of their respective labs, dedicated as they were to the task and supplied with everything they needed) understood that, as far as the general public were concerned, their research was both exciting and dangerous, because it touched upon questions that some considered forbidden territory.
Ever since Markram’s ‘Blue Brain’ project had successfully modelled a neocortical column, the question had been asked: When, if ever, would a virtual brain possess a virtual, conscious and self-aware mind? Could a simulation ever be said to be conscious, or was that something that no amount of calculations per second could ever capture? Early models had been impressive from the viewpoint of robotics, but less so from nature. Markram and his team had designed a robotic rat, remote controlled by a model of a rodent brain that existed within rows and rows of CPUs that collectively made up a supercomputer called Blue Gene. The robot rat was put through the kinds of experiments real rats were routinely asked to perform. Things like negotiating a maze. The real rats would always complete the task long before the robot rat. It was almost as if the latter were operating in slow motion — which indeed it was. Such was the complexity of running the simulated neocortex, that one second of thought and action required several minutes of number crunching. All but a few took this lag to be proof that a robot could never equal a living, breathing animal.
But, the model had been refined, and computing power had continued its exponential rise. It was not long before the robot rat was running through mazes every bit as quickly as its biological peers. And, it was not all that long before sales of real rats (and later cats, and later still, dogs) were falling as people saw the benefit of lifelike robot pets that would never die and cost a great deal less to upkeep.
Still, the perceived difference between animals and people meant that few dared to model a human to the level that many now thought would result in a conscious awareness. But, since people had been fascinated to know how the mind works since, well, forever, it was perhaps inevitable that, at some point, a research team would put together a system capable of growing a virtual baby that would become, in every respect, a person. Dr Giulio Dinova and his team had done just that.
The voice of Dr Dinova’s research assistant, Gwyneth Epsilon, awoke him from his semi-hypnotic state. The sensorium had the effect of making one loose sense of space and time in the physical world. Concentrating so much on the virtual world that Emily inhabited, one sometimes came to believe it was reality almost as completely as Emily herself believed. Dinova corrected himself. For Emily, it was not mere belief but a simple fact. What was it that the old roboticist Moravec had said? Oh yes, ‘to a simulated entity, the simulation IS reality and must be lived by its internal rules’. That…
“I said, has she been playing WeePeeple again?”. Dr Epsilon’s tone of voice suggested she had spent quite some time attempting to attract Dinova’s attention, and was becoming rather irritated at his absent-mindedness. His hand swept across empty space, and icons that seemed to hover in front of him dutifully scrolled from left to right. A finger jabbed at nothing, and Dr Dinova saw his finger touching the icon that minimized the Sensorium’s all-encompassing visual and audio rendering of Emily’s world. Moments later, a shimmering cloud had solidified into the shape of Dr Epsilon, accurate down to the last mole and laughter line. The two doctors were, physically, on opposite sides of the country. But in the age of augmented reality seamlessly blending the real and the virtual, people from around the world could collaborate as closely as any team whose members lived in close proximity.
“It’s funny”, commented Epsilon, in a tone that implied she meant ‘peculiar’ rather than ‘amusing’, “here is the most sophisticated artificial life in history, and yet she has to make do with a mouse-driven system. It’s like she is a late 21st century girl stuck with late 20th century technology”.
Dinova frowned. “You forget that our computing resources are not as deep as some suppose. Yes, yes, I know they exceed the entire computing capacity of the 20th Century Internet by an order of magnitude, but running a simulation of a person, right down to the synaptic firings and neurotransmitter concentration levels, is still a phenomenally intensive task. And then you have to take into consideration the fact that we have to render a physically-plausible environment for her. Of course, we are not simulating the world down to the level of particles — it is all tricks and sleights of hand designed to fool Emily’s brain into thinking it is inhabiting a physical place. But even so, we are close to pushing the limits of what is possible. I am afraid that emulations of crude videogames and their control schemes are just about all we can provide for decent entertainment. That, and dolls and cuddly toys that come with only the crudest models of child-parent social interaction”.
Epsilon understood all this. She also appreciated that it made the team’s job simpler. They were here to study, at a level of detail that had not been possible before, the stages of development that resulted in a newborn baby growing into a child with an inner life of her own. When she was playing with WeePeeple, Emily’s virtual mouse (which was solid and real to her, of course) permitted only two degrees of freedom. This restricted the amount of motor control her body needed to perform (she was mostly just sitting still, only moving the arm that controlled the mouse), which enabled the research team to follow the corresponding brain activity that underlined the building of interoception and exteroception maps of the internal state of her body, the world around her body and her body’s relation to the world. If she had been doing something like playing tennis, which involved using all of the body and required at least 19 degrees of freedom, the amount of data pouring from the computers that updated her brain and nervous system model would have far exceeded the team’s capacity to follow. They could only track such developments by relying on the Sensorium’s drastically simplified representations.
It was obvious that Emily enjoyed playing with Adam. And there was something fascinating about watching a virtual child, modelled so completely that she had a mind, guiding the development of a virtual person in a virtual world within a virtual computer that was itself within a virtual world, all ultimately existing as software within the rows and rows of supercomputers linked by super highspeed Internet3 links drawing on the spare computing cycles of the Cloud. And yet, Dr Epsilon felt slightly troubled. Her colleague Dr Dinova, his fascination with Emily and her daily routines seemed somehow detached and clinical. He reported on her progress as if he were a scientist noting the growth of some novel bacteria. But, for Epsilon, she felt more of an emotional attachment to Emily.
“Do you ever think about her 16th birthday?”.
Dr Dinova laughed. “Goodness, we have gathered so much data already, I think we will be spending five lifetimes, just in studying the details of development in the toddler stage! We will not be concerned with tracking the neurological underpinnings of morose teenagers until decades after the simulation is completed”.
Dr Epsilon shook her head. “No, that’s not what I meant. I meant, We are only running this simulation until our test subject is 16 years of age. After that, we shut everything down”.
Dr Dinova’s expression was one that suggested he failed to grasp the point. “Of course. Like I said, we are amassing such an overwhelming amount of data, we will have no choice but to halt the experiment at some point. I agree that any particular date is arbitrary, but a line must be drawn somewhere. Don’t worry, your job is not going to disappear as soon as we pull the plug. Like I said, we will all be engaged for decades to come, pouring through the data we have obtained during…”
“No, no”, interrupted Dr Epsilon, “that is not what I am getting at. I was just wondering about the ethics of it. Shutting down the simulation, won’t that be tantamount to murdering Emily?”
Dr Dinova looked serious. “Now look. Emily is not a little girl. She may resemble one, but she is not flesh and blood. She is a test-subject. She is an experiment. I know she has been designed to push evolutionary buttons that trigger the nurturing instinct, but never forget that she is, when all is said and done, nothing but a vast pattern of calculations”.
Dr Epsilon did not look convinced. “But she is everything we are, as far as we can tell. Her environment may be a crude approximation of reality, but she is not smoke-and mirrors. Her brain is a model that reproduces, in exact detail down to the molecular level, everything going on in my brain, or yours. Nothing but a vast pattern of calculations? What are we? Nothing more and nothing less”.
“We are getting into the realm of ivory-tower philosophy here”, countered Dr Dinova. “You must remember that we cannot determine if Emily really has a conscious mind. For all we truly know, she may be nothing more than a more convincing zombie than Adam. For that matter, for all you know I may be nothing more than a zombie too. Perhaps you are the only conscious entity in all existence. Maybe our reality is, itself, nothing but a grand illusion created by computers? We can argue about that until the end of time, and I dare say people probably will. But, there are empirical studies to be conducted, falsifiable theories to be tested. We have to remain impartial scientists first, and concerned parents of Emily, a distant second, IF we can permit ourselves to indulge in such roles at all”.
Dr Epsilon looked a bit sad. “I permit myself to feel like a guardian toward her. Of course, I do not actively make my presence known to her. I am not sure how her young mind would cope, knowing her reality is not real at all. As far as Emily is concerned, the memories we imprint are her actual parents”.
Like all people, Emily needed a social circle made up of family members in order to help her development. But, rather than waste computing cycles in simulating other virtual people to be her companions, the research team opted instead to imprint the memories of being cared for by a mummy and daddy. As far as Emily was concerned, her mother and father were in another room. Automated systems tracked her emotional state, and whenever she seemed to be in need of an adult presence, her simulation was paused, updated with memories of comfort and social interactions, before being allowed to progress. For Emily, her life was filled with interactions between loving parents, both of whom did not exist outside of her mind.
Dr Epsilon thought about those moments when Emily was on pause. It took a few minutes to insert the fake memories into her mind, during which time her awareness was zero. They were little deaths, these moments. And yet, whenever the simulation was unpaused, Emily would show no signs of understanding her world had been suspended. How could she know? Since she was not conscious during the moments when her simulation was paused, she perceived no loss of time. As far as Emily was concerned, her life ran continuously.
“I suppose”, said Dr Epsilon, speaking out loud to her colleague but really just voicing her thoughts to herself “that shutting her down for good is no more harmful than when we pause her simulation. She will not know she is dead, because she will not know anything. She will not be anything. Her mind, her body, her world, all of it will be nothing once the computers stop running through their calculations. Only…”
“What?”, asked Dr Dinova, looking genuinely curious.
“Well, do you remember that old roboticist? Professor Brezeal?”.
Dinova looked like he was wracking his brains. “God, you are going back a few decades. But, yes, I remember her. Never met her of course. She died when I was in my teens. But, now you mention her, Cynthia Brezeal’s work was one of the things that got me interested in this stuff in the first place. You know, she pioneered work in using studies from child psychology to build social robots. I mean, robots that could respond to, and give off, cues from body language, facial expression, and tone of voice, in order to establish an emotional connection with their users”.
Epsilon nodded. “Yes, that was her. One of Cynthia’s earliest creations was a robot head called Kismet. It was a terribly crude thing, inferior to the toys that Emily plays with, I would think. But, she became attached to it and she felt quite sad when the time came to leave Kismet behind. So, call me silly but I think there is no shame in admitting that I am not looking forward to the day when Emily is shut down for good. Not one bit”.
With that, Dr Epsilon shut down the communication link, and her avatar dematerialized in a puff of dispersing, virtual smoke. Dr Dinova sat in contemplation. He was thinking of the whole setup that allowed Emily to exist. All the supercomputers and the Internet3 links that sent information back and forth between them. A Web of supreme computing power weaving the magic of conjouring up a ghost-in-the-machine. He returned to the Sensorium, maximizing the window on Emily’s world, and observing from his godlike perspective, the detailed steps in her development.
