In The Singularity is Near, inventor and futurist Ray Kurzweil argues that the rate of technological progress, which is exponential, will reach a critical point about 2045 when its impact will be “so deep, that human life will be irreversibly transformed.” Before then, he predicts, human beings will have uploaded themselves to vastly more intelligent and robust machines. Kurzweil—who was born in 1948, as I was—expects to make that journey personally.
The Fog of Technology
One of many technologies Kurzweil mentions is called “Utility Fog.” The brain child of J. Storrs Hall, Utility Fog is made of foglets: “nanobots…that can manipulate image and sound waves [and] bring the morphing qualities of virtual reality to the real world.” [Kurzweil, 2005, p 28]
They’re called “foglets” because if there’s a sufficient density of them in an area, they can control sound and light to form variable sounds and images. They are essentially creating virtual-reality environments externally (that is, in the physical world) rather than internally (in the nervous system). Using them a person can modify his body or his environment, though some of these changes will actually be illusions, since the foglets can control sound and images. Hall’s foglets are one conceptual design for creating real morphable bodies to compete with those in virtual reality. [Kurzweil, 2005, p 310]
Foglets are miniature computers which, linked in a vast network, would have essentially unlimited processing power. They are equipped with telescoping arms that can join ‘hands,’ allowing them to form solid shapes, or behave as liquids or even gases. Microscopic in size, foglets control the colour of light they reflect by manipulating an ‘antenna arm’, giving them the ability to take on the appearance of any material. And, Hall wrote…
…since you’re embedded in the Fog, it can sense every detail of your bodily position. It forms a “whole-body dataglove,” and you can control it with extremely subtle gestures. At the ultimate extreme, the Foglets can carry various special sensors ranging from simple electrodes with voltmeters to SQIDs and form an extremely high bandwidth polygraph. With proper programming the Fog would almost be able to read your mind. [Hall, 2001]
Reading this, I was reminded of a novel I wrote in my 20’s, now in a desk drawer (where it will remain). It was a ‘big-idea’ dystopian vision of technology pushed to its limits. Technology exists to satisfy human desires with a minimum of time and effort. The ultimate technology would be one that responds directly to people’s desires, by interpreting their brain-waves, and reshapes their surrounding physical environment into the image in their mind’s eye. Hall, I realized, had leapfrogged my literary vision by sketching out a physical implementation.
The fog gives people what they want; but not everyone wants the same thing. My novel had to deal with how to resolve different people’s conflicting desires. That’s not terribly difficult, it turns out, for a technology that shapes an infinitely plastic world. The solution is localization. If my wife wants to paint the livingroom walls white, but I prefer green, we can both have what we want (and save energy at the same time!) if the colour changes happen, not on the physical walls, but on surfaces close to our eyes, like the binocular display of a virtual-reality helmet which would always show the actual location of the walls to both me and my wife (so that we wouldn’t bump into them), but which would show each of us the walls in the colour we prefer. And because the display is not actually worn, like a VR helmet, but cleverly suspended wherever we happen to be looking at the moment, we need not be aware that it is so very local; the experience would be just like looking at the real walls of the room. Or like looking at the walls of a ‘room’ that exists only in a Matrix-like virtual reality. And it would be hard to tell the difference between virtual and real.
This is exactly the sort of thing Hall and Kurzweil have in mind. “Foglets…can direct visual and auditory information in such as way as to bring the morphing qualities of virtual reality into real reality.” [Kurzweil, 2005, p 33]
Fog-like technology enables a kind of isolation, a privatization of experience in which each person inhabits a local reality of his or her own wish-fulfillment, largely or entirely unshared with other people. Living in a Utility Fog, even our encounters with others would be heavily mediated.
Other people (such as your romantic partner) will be able to select a different body for you than you might select for yourself (and vice versa). [Kurzweil, 2005, p 29]
Then why would you bother with a human—occasionally cranky and uncooperative—romantic partner at all, when you can shape foglets into the girl or guy of your dreams? You needn’t even know you’re doing that. Utility fog could respond to subconscious as well as conscious brain activity and build you a lover you would have no reason to think wasn’t real. You could easily persuade yourself she had a mind of her own if that’s what you wanted to believe.
And so Utility Fog is revealed as the ultimate tool of self-deception.
Whether we are deceived about it or not, immersion in a technology-doped environment that obeys our every voiced and unvoiced whim has the potential to eliminate all frustration from human experience. Real adversity and challenge are banished from this world; all that remains is fake adversity, the illusion of challenge tamed to the comfort zone of the participant. And therein—here comes the Big Moral of the novel I wrote as a young man—lies the weakness at the heart of technological progress. Genuine adversity and challenge are needed to shape us. Foglets would allow people to win every race they entered, and every poker game too. Amateurs climbing Everest would never die (although they might ‘see’ other climbers die). Every wannabe celebrity would be one—and would never know he was famous only to a single fan. High honours and office need not be hard won. The price of localized wish fulfillment is an inevitable loss of integrity—not just moral, but psychological and physical integrity. Immersed in a utility fog, human beings cannot help but drown.
Uploading to Immortality
Foglets are mentioned only in a few scattered paragraphs of Kurzweil’s book. A more central theme is the looming merger of human minds and machine intelligence, which Kurzweil predicts will take several forms:
- Enhancement of the human body with embedded machines (high-res artificial eyes and ears, chips to extend the range and accuracy of human memory, armies of nanobots in the bloodstream repairing damage and reversing biological aging).
- Uploading human minds to computers controlling robotic bodies with superhuman capabilities. Consisting entirely of replaceable parts, and provided with safe, out-of-body storage of their memories and psychological states, these beings will be effectively immortal.
- Dispensing with physical robot bodies in favour of life in a Matrix-like computer-generated virtual reality. In this scenario, people could, Kurzweil supposes, remain individual persons with distinct lives, psychologies, and trajectories through virtual space. They could still be moral agents, responsible for their own actions.
- Dispensing with individual lives, in a transcendent fusion with one another and the host machine.
In Kurzweil’s eyes, these are not just dreams of a far-off future. His stake in his predictions is a personal one. Kurzweil plans to participate in these stages of progress—all of them—himself. Kurzweil openly admits to being motivated by personal survival and his desire for immortality. He is puzzled by others, of his own age cohort, who are not excited by his vision—the tantalizing nearness of eternal life.
We have the means right now to live long enough to live forever. Existing knowledge can be aggressively applied to dramatically show down aging processes so we can still be in vital health when the more radical life-extending therapies from biotechnology and nanotechnology become available. But most baby boomers won’t make it because they are unaware of the accelerating aging processes in their bodies and the opportunity to intervene. [Kurzweil, 2005, p 371]
Kurzweil’s forecasts of technological progress leading to the Singularity, by about 2045, are based on the same well-documented, remarkably stable, exponential trends—Moore’s Law and related ‘laws’—that I used to support the claim that replication of living biological human beings could become a reality by 2058. It’s unlikely that both our predictions will be correct. If Kurzweil’s vision is closer to the truth—people abandon their limited biological bodies for artificial bodies with more processing power and replaceable parts—then the future I envisaged will be overtaken by his. Why would a society of intelligent robots bother with scanning and duplicating their physical bodies? Teleportation would take a different form: as a robot in 2030 wanting a vacation in French Polynesia (assuming I would still want it) I would simply order up a robotic body (maybe the same model as my own, maybe an upgrade, maybe a rental, maybe one customized to look like me) and transfer my full psychological state, including memories, personality, and vacation plans, to it. Medical and cosmetic treatments would not require full-body scanning and manufacturing; defective parts would simply be replaced, as would parts that did not please their owners aesthetically. For variety, we might own several robotic bodies (possibly stored in different places) of different sizes, shapes and sexes, which we could activate serially like changing our wardrobes. Life insurance would take the straightforward form of making frequent—perhaps continuously updated—backups of our memories and other psychological attributes to secure cloud storage.
But the devil is in the details, and the details of how the future actually turns out frequently bedevil futurists. Kurzweil is fallible on this score. The Singularity is Near, published in 2005, includes predictions for the 201x’s which—now that we have arrived at that vantage point—are significantly off.
Early in the second decade of this century, the Web will provide full immersion visual-auditory virtual reality with images written directly to our retinas from our eyeglasses and lenses and very high-bandwidth wireless internet access woven in our clothing. These capabilities will not be restricted just to the privileged. Just like cell phones, by the time they work well they will be everywhere. [Kurzweil, 2005, p 472]
In January 2013, eyeglass displays and internet-enabled underwear remain—as they were in 2005—exotic. At least, I don’t know anyone who owns them. As for full-immersion visual-auditory virtual reality on the Web, Second Life (started in 2003) is still around, and it has improved, but it hasn’t been adopted at anything like the pace of, say, the iPhone, (first released in 2007). Looking for the most transformative technological changes in the eight years since Singularity came out, I’d single out smart phones and social networking. Wikipedia (a transformative technology in its own right) tells me today, “As of September 2012, Facebook has over one billion active users, more than half of them using Facebook on a mobile device.” (and helpfully cites its sources, links to which I have pasted here). Eyeglass displays and networked clothes? Not so much.
And that’s instructive. The inseparability of humans from their phones is a familiar joke. The sight of people ignoring each other at a party while they interact with their hand-held devices barely registers—although it can still amuse us if we take time to notice it. (It’s the subject of several New Yorker cartoons.) Smart phones have changed social and political reality on all scales, from allowing the shy to avoid scary real-time encounters with people in the flesh, to enabling coordinated political actions that topple governments. Perhaps the Arab Spring would have happened without cell phones—but definitely not in the same way. Having Google always at our fingertips is another profound change. At a party last month, someone said, “Remember what it was like when we wondered about things?” Now we ask Siri. Functionally, that’s not unlike having an internet connection implanted in our brains.
I also wonder whether life on Facebook (launched in 2004), where a lot of people spend their spare time these days, is a harbinger of life in virtual reality. The popularity of social networking is clearly another transformative change. We are accustoming ourselves to interacting in a kind of virtual space which is structurally different from the real space we were born into—more different than the virtual-reality spaces described by Kurzweil or depicted in the Matrix movies. Your personal ‘Facebook space’ is defined by your friends list and your (nebulous, confusing, unstable) privacy settings. We read, respond (‘like’), and post into a scrolling list of messages that overlaps with other people’s scrolling list of messages which they may or may not ever read. We have a sense of being in a ‘room’ with other people, as if at a party. Some of these people we know well, others slightly, some by reputation only. Some of them we may not have a clue about. We (usually) can’t see who is in the ‘room’ when we ‘say’ something, and we can never be sure which of our ‘friends’ will later stroll into the ‘room’ and read what we wrote before it scrolls down into oblivion. But actually there is no one ‘room’ in which we interact, in this odd, time-delayed way—everybody at the party is in his own ‘room,’ with his own set of ‘friends’ (some of whom are present, others not), and all those rooms partially overlap! If you and Ray are both in my room, it doesn’t follow that you and Ray are in a room together; and I will assume you are not unless I happen to know that you and Ray are ‘friends.’ Despite the wildly creative graphics and teleportation portals in the world of Second Life, Facebook space is a stranger place—yet people take to it like ducks to water. Second Life and ‘classic’ virtual reality are attempts to simulate ‘natural’ 3D space in cyberspace; but Facebook space takes its shape from the way its users choose to interact, within the range of possibilities enabled by the technology. The ways in which social networking has evolved over the past nine years were not predicted by Kurzweil or anyone else. The example shows that, when we futurists try to imagine how life will change as human beings increasingly integrate themselves with their own artifacts, we should not bet too heavily on getting the details right. I don’t mean that thinking seriously about the future is not worthwhile. We need to speculate, quite liberally, in order to foresee dangers before we stumble into them, if we are to have any hope of producing an actual future which is better than most of the possible futures now open to us.
Ray Kurzweil’s predictions of uploading may or may not be borne out. If they are, his predicted schedule may be off. Predictable exponential improvements do not determine the direction of technology, or the actual devices that will be built. Those devices will be determined by the swirls and eddies of money through the marketplace, propelled in turn by the choices made by billions of people. Particular swirls and eddies—like the weather on today’s date a year from now—are beyond our powers of prediction. Personal uploading to machines, in the forms Kurzweil envisages, may not actually happen. But even if it does not, it is worth discussing, because that will help people make more intelligent choices.
Can Ray Kurzweil Survive the Singularity?
Readers of the Phantom Self may be familiar with my views about survival: that although people are strongly inclined to believe that there is a fact of the matter as to whether or not they survive any given transformation (such as information-based teleportation, catastrophic dementia with personality change, or just waking up tomorrow morning), there is no such fact; no scientific test can ever settle the question of whether or not you will survive. What we have instead are two sets of facts, loosely related: facts about continuity and connectedness between persons, and facts about attitudes towards persons. Except for a handful of philosophers, everyone thinks that he will survive a contemplated transformation if, and only if, he has the same sort of motivating attitude, or concern, for the survivor of the transformation as he normally has for himself. But people do not agree about the relationship between these two sets of facts; some think they would survive teleportation and others do not; some, diagnosed with severe rapid-onset dementia, personally dread the fate of the survivor, whereas others adopt an attitude that is closer to the sympathetic concern they would have for any dementia patient; and (most pointedly and remarkably) many people expect to survive their biological deaths either in a disembodied state, or reincarnated in another terrestrial organism which might or might not be human. In addition to this tremendous variation between people regarding the transformations they expect to survive, thought experiments reveal that the psychological connection each individual forges between the two sets of facts is plastic. Different descriptions of a single set of facts can pull the same person in two different directions on the question whether or not he himself will survive a given transformation. But there’s no ‘right answer,’ because personal survival is not a fact—it’s an attitude.
Ray Kurzweil takes his stand far out on the ‘liberal’ end of the spectrum of personal identity theories. He clearly expects to survive radical transformations that other people would regard as no better, for themselves, than dying. He is strongly motivated to transform himself, and he expects to have the experiences of the survivors of those transformations—he expects to “be around” to “see for himself” what post-Singularity life is like. He supports this philosophical position with the idea that persons are ‘profound patterns.’
It is not unreasonable to regard patterns as a fundamental ontological reality. [Kurzweil, 2005, p 478]
I suspect Kurzweil would look favourably on my theory that persons are best regarded as informational entities.
Kurzweil holds his position on survival somewhat uncritically. I say that because he clearly thinks that (1) there is a crucial difference between an outcome in which he survives, experiencing the Singularity and post-Singularity events, and one in which he does not, yet (2) he makes no serious attempt to define the boundary conditions which make that pivotal difference.
Even if the future unfolds, in broad strokes, as Kurzweil expects, he may not in fact ‘live to see it’ in the way he hopes to. Despite taking good care of his health, and consuming 250 ‘nutritional supplements’ daily[Kurzweil, 2005, p 211], he may yet suffer a fatal cycling accident or medical emergency before he has the opportunity to be uploaded. But even in that unfortunate event, much of Ray Kurzweil would survive: his books and videos, to begin with, which are very influential, propagating his ideas—‘profound patterns’ of thought—to many other minds, some of which belong to young people who will probably be around to see the future of 2045. Kurzweil obviously doesn’t think that dissemination of his ideas is enough for him to survive; but we must challenge him with the question, “Why not?” Perhaps he would reply that too much would be lost; his life is more than his books; so many of his memories would be lost; personal relationships with other people would be broken off; ongoing projects would be aborted; and so on. That reply seems reasonable enough, but it invites the question why too much would not be lost if he were uploaded to a robotic body or to life in virtual reality. Either of those transformations seems to involve profound changes, which—someone might argue—are likely to disrupt the ‘profound pattern’ which is Ray Kurzweil, perhaps beyond recognition. Even his own.
Of those two transformations, uploading to a robotic body with an artificial brain is the less radical. Kurzweil argues that we have already embarked on this journey. Titanium replacement joints are commonplace; and increasingly sophisticated mechanical prosthetics largely overcome the disability of amputated limbs. New Brain-Computer Interfaces (BCI’s) have been developed allowing completely paralyzed people—who until recently would have experienced the hell of being “locked in,” conscious but unable to act or communicate—to use a robotic arm to feed themselves, and write their thoughts (albeit slowly, so far) on a computer screen. [Wikipedia, “Brain-Computer Interface”] BCI’s that respond to electrical signals in the brain effectively dissolve the most important barrier between human and machine, allowing us to use computer memory to supplement human memory without the cumbersome indirection of having to type or talk. Although ‘external’ to his organic body, the computer and robotic appendages which a locked-in patient can control function as part of his body, in that they allow him to be an agent, to act.
Kurzweil tries to make the question whether he would survive uploading more objective by proposing a ‘personal Turing Test.’ [Kurzweil, 2005, p 200] In the original Turing Test for machine intelligence, human judges have conversations over text-only channels with subjects who may be human or computers programmed to imitate humans. If the judges cannot reliably distinguish the humans from the computers in the test sample, the machines are considered to have passed the test. Kurzweil describes the ‘Ray Kurzweil Turing Test’ in similar terms. If human judges who know Kurzweil personally cannot tell the robotic version of Kurzweil from the original biological one by Q&A texting, the robotic version passes the test and is—in every way that matters—Ray Kurzweil.
I’m not convinced that the robotic Kurzweil would be able to pass the ‘Ray Kurzweil Turing Test.’ Oddly enough, I think it’s more likely he would pass if the test were administered soon enough after Kurzweil was ‘uploaded,’ than if a year or two had gone by and he had adjusted to robotic life. Life with a robotic body may be expected to be quite different from life with a human body, and that difference will leave traces which a clever human interlocutor who knew Kurzweil well should be able to detect, and which the robotic Kurzweil, extremely clever though he be, will find it hard to mask.
My reasons for skepticism are based on probable aspects of non-humanness in robotic life. The whole point of trading in our biological bodies for artificial ones is to overcome shortcomings of the former. The decay of aging, the limit of death, for a start, affect the quality of human life deeply; once we are uploaded to robotic bodies, these constraints will be eliminated. Ray Kurzweil obviously worries about his own health and the possibility of dying; that’s why he invests so much time, energy and money in maintaining his body in as youthful a condition as possible. But once he is a robot, he will be able to breathe a sigh of relief and forget about all that. No more nutritional supplements! No more regular checkups with Dr. Terry Grossman. A major focus of Kurzweil’s life would disappear—replaced, no doubt, by more interesting pursuits. But once out of mind, aging and mortality are subjects on which Kurzweil’s Turing Test responses are likely to change. After two to five years of robotic life, they would probably change a lot. Of course, the robotic Kurzweil, given the mission of trying to convince the human judge that the person she is texting is the organic Kurzweil, would do his best to channel his former biological self. But would he succeed in convincing a qualified judge, who knew the biological Kurzweil well—a close friend, or his wife Sonja (a child psychologist)? It would be rather like trying to impersonate yourself at a much younger age, maybe eight. Could you do that well enough to fool, say, your mother? I doubt I could. I’d even have trouble simulating my twenty-five-year-old self convincingly. But the changes in life-style, and the psychological changes which, it seems reasonable to expect, would accompany them following uploading to a robotic body vastly exceed the changes in my life between the ages of twenty-five and sixty-four.
As robots, we would never get tired (let alone cranky). It seems unlikely we would need to sleep. Not sleeping, we would have no more dreams. So if, during our personal Turing test, our judge asked what dreams we had had lately, we’d be stuck for an answer—or we’d have to make something up. But real dreams are so weird, they’re hard to make up convincingly! Maybe Kurzweil would remember dreams from his biological life, and could relate those. But the judge—if given the opportunity to question Kurzweil on several occasions—might smell a rat if there was never any new material. And she could go on to ask how Kurzweil feels at the end of a hard day’s work. Does he feel the urge to put his feet up? Pour a glass of single malt? Does he still get creative ideas, or do those come when he wakes up in the morning? To pass the test, the robotic Kurzweil will have to prevaricate extensively. To convincingly sustain a fabric of lies under interrogation—if you have ever tried it—is damned difficult.
Maybe robot Ray can pull it off—he’ll have a better memory, and more cognitive resources, than you or I. Still, the need to lie so much in order to pass the test is, in its way, a measure of how different from an organic human being Kurzweil will have become.
Not convinced? Then consider a soldier with PTSD, back from a harrowing tour of duty in a campaign like Iraq or Afghanistan, nerves wrecked by too many close encounters with IED’s. A guy—call him Joe—who breaks into a cold sweat walking down a crowded street, or wakes up screaming when the night wind bangs a gate. What will happen to Joe when he is uploaded to a robot body? Clearly, some changes are in order; neither he nor his family would want his PTSD to be programmed in. And so, although Joe will remember everything, he will lose his acquired neurosis. But that should make it easy for the military psychiatrist who judges his personal Turing Test to tell the difference between the biological Joe and the robotic one—just see how he reacts to the kinds of stimuli that trigger his PTSD symptoms.
When I imagine myself and the people I know best—members of my own family—making the transition to robotic bodies, one salient difference stands out—our relationship to food. It’s improbable that artificial bodies will get their calories from organic food. But will they eat? Kurzweil describes them as having sex, although it would play no reproductive role. Kurzweil is undoubtedly right on that point. Most people would balk at being uploaded to robotic bodies if it meant giving up sex forever. The robots would be fitted out with the appropriate organs and secondary sexual characteristics to have a grand old time; they would be athletic enough to work through all the positions in the Kama Sutra and then some; and there is every reason to think that their experience would be undiminished, and probably enhanced, given that enhancement of sexual pleasure (as we know from the supermarket magazine rack) is a prime motivator for a lot of people. But food? Kurzweil—perhaps betraying the indifference of a man who subsists mainly on pills—is silent on the subject.
In my family, food is a frequent topic of conversation and focus of activity. “Frequent” hardly does justice to it; I’ll go further and say that we hardly ever get together without a conversation about food—its preparation, recipes and cooking techniques and equipment; reminiscences of past meals, good and bad (almost all good); restaurants we have visited where we have enjoyed dishes we later tried to emulate; and so on, at length and in depth. Any member of my immediate family would think twice, or thrice, about uploading to a robotic body if it meant giving up eating.
At this point, I feel compelled to drop into the dialogue narrative style which Kurzweil scatters through The Singularity is Near.
Ray (2004): You won’t have to give up food. You’ll be able to enjoy virtual food in virtual reality.
Gordon (2013): That seems like thin stuff. And where’s the satisfaction in cooking it, the challenge of producing something truly delicious? Anyway, we’re discussing robotic bodies in the real world, not virtual reality yet.
Ray (2030): Lot of my robotic friends eat food. They’re equipped with olfactory and gustatory sensors that deliver more information to their brains than your aging palate. Satisfaction too, I should add (though I’m not into it myself).
Claudia (Gordon’s wife, 2013): What happens to the food they eat?
Ray (2030): It gets chewed up and swallowed.
Claudia (2013): And when their stomachs fill up?
Ray (2030): It’s removed and composted.
Claudia (2013): That’s obscene! If there’s one thing I can’t stand, it’s a waste of food.
Gordon (2013): It’s true. I can’t see you doing that.
For Claudia, a strictly recreational relationship to eating would not work, because of her deeply-ingrained revulsion to wasting food. I’m reasonably sure that, as a robot, she would choose to turn her appetite switch to the Off position. If she took the Claudia Turing Test, judged by either of our children, she would inevitably fail—not because she’s a bad liar (although she is) but because they could easily tell that she no longer had her characteristic lively, immediate, passionate interest in food. Without that, would she still be the same person? (For my admittedly complicated and counterintuitive answer, refer to the first paragraph in this section, following the header Can Ray Kurzweil Survive the Singularity?)
One more word:
Gordon (2013): Relax, Ray. Don’t worry so much about whether your predictions will work out or that your biological body may die before they come true. You have succeeded in spreading your ideas and inventions far and wide, influencing the lives of many people—including a blind friend of mine who, in the late 1970’s, acquired a Kurzweil Reading Machine which hugely improved his productive capacity, his employment prospects, and his happiness. You have succeeded in staking a huge claim on the future, which is all that any of us can hope for. You have already uploaded.
Hall, J. Storrs (2001) “Utility Fog: The Stuff that Dreams Are Made Of,” http://www.kurzweilai.net/utility-fog-the-stuff-that-dreams-are-made-of, accessed Dec. 18, 2012
Kurzweil, Ray (2005) The Singularity is Near, Viking.
Wikipedia, “Brain-Computer Interface,” accessed Jan. 2, 2013