..

Chapter 1 - The Extraordinary Future

Author: Winfred Phillips

The average human life span is less than a hundred years. If we are lucky enough to make it that far, our bodies are so worn out that our earthly lives soon end. Many of us succumb to some disease or other and don't make it that far.

Some religions or their followers make claims about life after death, but of course in the case of any particular religion or its claims there are other people who don't believe those claims are true. Actually, some people believe that all such religious claims about life after death are false or dubious. There don't seem to be any religious claims to immortality that are accepted universally. So it's not uncontroversially true that there is life after death.

We're living in the computer age now, but so far that fact has seemed irrelevant to the question of life after death. Sure, computers may have some place in medical technology, but they won't enable us to live forever.

Some computer scientists say this latter claim is false--that in the next century computers will enable us to live forever! We won't have to wait for our bodies to degenerate to the point where they bring about death. Prior to such a point, each of us will be able to transfer his or her mind to a robot and continue living in the robot body.

In this thesis I will examine the plausibility of the claim that mind transfer from a human to a computer (robot) will occur before the end of the twenty-first century. I seek to determine to what extent we can be confident that mind transfer will occur during the next century. To do so I will examine relevant technological issues as well as philosophical issues pertinent to this claim. If a statement is needed of the "thesis" that I will defend, it is that the authors who depict the near arrival of human-computer mind transfer are too optimistic in their prediction.

But let me say at the start that my perspective will be both sympathetic and critical. Several authors have developed a general line of argument supporting the claim of the impending possibility of human-computer mind transfer. I will examine this line of argument critically to see if it really sounds plausible. But where it does not, I will sometimes try to explore options for shoring it up. I have no hidden political or religious agenda in support of or against human-computer mind transfer. All I want to know is to what extent we can have confidence that such a thing could happen in the next century. This thesis will argue that we should not be confident.

I do not know whether you wish to consider the prospects for human-computer mind transfer--perhaps you think the claim is so preposterous that you have dismissed the possibility of even considering it. I do not think we can know that it will never happen. Perhaps you will be interested in at least considering the question after reading the claims of those who think it will happen. Even if you don't care about human-computer mind transfer, you might nevertheless be interested in the technical and philosophical issues discussed in my investigation.

The claim is that a human being will be able to transfer his or her mind to some kind of smart or potentially smart robot and continue existing as the same person but now in the robot brain and body instead of the old human brain and body. If we step back from the claim we can see that a number of further issues and questions arise immediately. Is the human mind such that it could be transferred? What is the human mind anyway--is it the brain or something more? By what technical means could we get that mind out of a human brain and body so to transfer it? By what technical means could we put it in a computer or robot? What would the robot have to be like in order to "receive" that mind? And after the purported transfer, would it really be the same person in the robot body that was in the human body? In other words, would it really be the transfer of me (my self), rather than the creation of some new person?

With so many issues and questions, we could start in any number of places. I choose to proceed in somewhat the same order that most authors have adopted in arguing for the claim. They don't first get involved in a complicated discussion of the human mind and whether and how it could be captured and transferred. They don't start by discussing the issue of personal identity. Rather they start by trying to convince you that robots in the twentieth century will get really smart, as smart or smarter than humans.

I can think of two reasons they start here. First, they want to get you excited about robots. If they can convince you that robots in the twenty-first century will be really smart, smarter even than humans, then you will be more enthusiastic about the technological possibilities and perhaps see claims about human-computer mind transfer as not so far-fetched after all.

A second reason for starting here is that the authors who argue for the impending reality of human-computer mind transfer often have other matters on their agendas. Not everyone who writes about it is discussing just this subject. These writers vary in the extent to which human-computer mind transfer is seen as a major theme in their depictions of life in future centuries. Their books are not solely about human-computer mind transfer but also about how advanced computers and robots will soon be. They think robots will be so advanced that some will dwell in society much like humans--walking and talking, living their own lives, etc. In future centuries robots will even colonize space. Discussions about human-computer mind transfer often occur in the midst of this kind of advancement of a larger robot program, so the natural place to begin their discussions is with the robots themselves.

How does human-computer mind transfer fit into the robot program? As it turns out, humans might miss most of the "action" if they remain limited to their current bodies and capabilities. We are not nearly as smart as smart beings could be, and our bodies are not all that capable or durable. If we don't want to miss most of the action, we really need to find a way to overcome the limitations of our human bodies--we need to get smarter, stronger, and more durable. Human-computer mind transfer is seen as the way to do that.

In this thesis I am not considering the larger question of the future of robots in general, nor am I trying to get you excited about robots, so I don't have to discuss all the claims and arguments about robots that occur in the writings of the authors we will examine. Neither do I have to follow their order of exposition, but in explaining their position on human-computer mind transfer it will turn out to be convenient to start with robots. A lot of the technical issues that come out in the robot discussion will be useful later. So first I will consider the question of where robots will be in the next century. Then I will examine the issue of what the mind is and whether it is the sort of thing a robot could have. I will next consider possible mechanisms by which a transfer could take place. Finally, I will discuss the issue of personal identity--will it be the same person after the transfer?

I do think that for a human person to transfer his or her mind to a computer and continue personal existence in a robot body, the following predictions must be correct:

1.     Robots will be as smart as humans are.

2.     Robots will be capable of being persons.

3.     There will be a viable mechanism for the transfer of a human's mind from a human body to a robot body.

4.     Transferring one's mind to a robot will allow one to continue one's existence as the robot.

For human-computer mind transfer to occur in the twenty-first century, as is being predicted, such events must happen within the next hundred years. But consider why these events must occur at all for human-computer mind transfer to work. The basic idea is to put your mind into a robot brain, which will be some kind of compact computer. If the computer is just not capable of supporting your level of intelligence, then such a transfer is undesirable and it's not plausible to consider it a transfer of your "whole" mind and therefore it may not even be a transfer of "you." The basic idea is to maintain or improve your lot, not turn you into a dolt that has little claim to being you anymore. So computers, or the robots containing them, will have to be at least as smart as humans. But being a human person involves more than just being intelligent--it involves being conscious and having other features of personhood, such as exercising a free-will, or at least thinking we do, and being able to act morally. We can't lose these features in the transfer if we are to remain persons, so another requirement is that robots must be capable of supporting personhood, or being persons. In addition, of course, there must be a viable mechanism of transfer for it to work. Finally, part of the inspiration for being interested in human-computer mind transfer is to gain immortality. It would certainly seem to be the case that for this to happen the robot-being existing after the transfer must be the same person as the one who went into the transfer (as the human). So it seems that one's personal identity would have to be maintained--transferring my mind to a robot will allow me to continue my existence as the robot.

The authors who write about such matters think that not only will human-computer mind transfer be possible in the next century, it will actually happen. I'm not going to worry about this distinction between it being able to happen and it actually happening. I'm not interested in the question of whether if it might be technically possible to pull it off, say in 2090, political and economic factors will allow it to really happen or instead prevent it from happening in anything more than research labs. If during the next century we have computers capable enough to allow it to happen, that's good enough for our purposes, whether or not political and social factors allow it to become a widespread phenomenon in society. We're interested in the possibility and will assume that if it is possible for it to happen, in the sense that the robots and transfer technology are ready and other technical and philosophical questions have been resolved, then that's enough to make the claims about it happening plausible.

Let me make some comments on terminology here. We are mainly interested in robots, which might be considered embodied computers. The general idea is that the robot will have some sort of computer as the equivalent of a human brain, with some kind of body incorporating different types of sensory apparatus. The body need not be made of metal; as we will see, new technologies may involve other materials. (However I will not consider a being whose body is an exact duplicate of a human, made of human tissue, to be a robot. The idea behind human-computer mind transfer involves more than just making a duplicate human being.) I will use the term "robot" most often but sometimes I will use the term "computer" when I mean the computer in the robot or the robot itself.

"Robot" is from the Czech word "robota," referring to a peasant or someone engaged in forced labor. "Robot" was first used in a 1920 play to refer to mechanical creatures who could carry out routine tasks without much external instruction. Thus the original use of the term refers to an "intelligent" machine that is mostly self-controlled. Associated with "robot" are the terms "android," which refers to a bi-pedal robot that looks human, and "cyborg," which is a combination of human and machine (Paul & Cox, 1996, p. 25). Mostly I just use the term "robot," though it may be that the robots in question turn out to be androids. On some human-computer transfer scenarios involving transplants as part of the mechanism, the intermediate beings might be considered cyborgs.

I also use "computer" in the sense most people mean, for modern digital computers. In the future they might be made of different materials. On my usual use of "computer," a human brain is not a computer. But some authors use the term in an extended sense for "any system that uses signals to process information via calculations that solve algorithms according to a set of rules (Paul & Cox, 1996, p. 38)." In this extended sense of "computer" it may be that the human brain is a computer. I do not use "computer" in the sense Searle (1981) does, in which everything can be considered a computer because it can be interpreted as instantiating some function or algorithm.

Sources Describing the Extraordinary Future

The basic position this thesis will discuss holds that within the next century computers and robots will be at least as smart as humans, robots will be conscious persons with synthetic bodies, and human beings will be able to continue their personal existence by transferring their minds to such robots. This may seem a startling claim more at home in science fiction than science, but to scientists holding this view the position is considered between the two in the realm of "science speculation." I will borrow a phrase from Paul and Cox (1996) and henceforth refer to the happening of this series of events as the "extraordinary future."

Apparently many computer scientists and private industry computer executives working in artificial intelligence believe that eventually we will develop robots as smart if not smarter than human beings. This view is not usually discussed in scientific journals but comes out in interviews, popular articles, etc. Of course, not all of these scientists and executives believe, as the authors we will examine do, that this will happen during the next century.

Some of the scientists who believe in the upcoming era of robot intelligence also hold that such robots will be conscious persons. Probably still fewer believe that we will be able to transfer our minds and identities to such robots, but among those who do are a few who make this prediction in some of their popular writings. For my description of the extraordinary future I rely on Ray Kurzweil, Hans Moravec, Gregory Paul, and Earl Cox. Kurzweil, founder and executive of several companies involved in pattern recognition and other artificial intelligence, presents this idea in his recent work, The Age of Spiritual Machines: When Computers Exceed Human Intelligence (1999). Moravec, founder of and now Principal Research Scientist at the robotics program at Carnegie Mellon University, presents this position most recently in Robot: Mere Machine to Transcendent Mind (1999) but also earlier in Mind Children: The Future of Robot and Human Intelligence (1988). Paul, an evolutionary biologist, and Cox, a computer industry executive, present their ideas in Beyond Humanity: Cyberevolution and Future Minds (1996). Throughout this thesis I will use the phrase "our authors" or "the authors" to refer collectively to this group of individuals.

For my depiction of the extraordinary future, I really will draw on just the above authors. But as mentioned, others have similar opinions. I will not draw on the comments of these other thinkers because it might not be considered fair to critically analyze a position on the basis of some casual comments made in a brief interview, for example. But the reader interested in hearing such comments should consult Fjermedal's The Tomorrow Makers (1986). Fjermedal elicits a variety of comments sympathetic to the extraordinary future from distinguished computer scientists. Moravec is quoted extensively. But there are others. For example, Marvin Minsky apparently believes human-computer mind transfer will happen, though not in his lifetime. Danny Hillis sees it coming, as does Gerald Sussman, who thinks it could happen in the next generation (Fjermedal, 1986, pp. 7-8). Such thinkers are well known in the computer field.

In each of the following chapters of this thesis I examine one of the basic predictions I laid out above, namely, that robots will be as smart as humans, that they will capable of being persons, that there will be a viable mind transfer mechanism, and that human-computer mind transfer will allow maintenance of personal identity. In each chapter I proceed first by describing the pertinent views of our authors, mentioned above, after which I bring in supplementary material and critically discuss the plausibility of the prediction coming true. It may help you to follow what is going on later if at the start here I lay out the extraordinary future in a little more detail, so below I flesh out my earlier description. Discussion in more extensive detail and citation of specific sources will have to wait until later chapters.

Description of the Extraordinary Future

Our authors think that in the next century computers will get very smart. The reason is that computing power will continue to increase exponentially. Within the next century computing power available in a compact package will first equal the computing power of the human brain and then far surpass it.

Several approaches are used to estimate the computing power of the human brain. One method is to try to base it on the number of neurons, connections, and firings in the brain. The human brain has roughly on the order of a hundred billion neurons, with each neuron connected to as many as thousands of others. Each of these connections is a synapse, so there are trillions of synapses. An estimate is then made based on taking each of these connections to represent a calculation and noting the number of times neurons can fire a second. Similarly the number of brain connections is used to estimate the memory capacity of the brain. Another method for estimating the computing power of the brain is to first estimate how much computing power it takes a digital computer to carry out a particular brain function, and from this extrapolate to how much computing power it would take to do all the brain functions.

Estimates for brain computing power vary among our authors, but typically it winds up being in the hundreds of trillions or quadrillions of calculations or instructions per second, and memory is estimated at about the same number of bits.

Our authors also note that the brain appears to work by some kind of massive parallel processing rather than by the kind of serial or sequential processing typical of modern digital computers. While in the future computers will involve more parallel processing than they do now, they should not have to match the massive parallelism of the human brain. This is because they can make up for less parallelism by running faster than the brain can.

Though modern computers are far from possessing this level of processing power now, computers in compact form should attain the requisite processing power sometime around 2020 to 2040. This is because, as Moore showed in 1965, processing power had to that point doubled every year. Moore also claimed that this trend would continue; computing power would grow at an exponential rate, doubling every one to two years. The claim that it will has become known as "Moore's Law." The authors think that this exponential rate of growth can be sustained by the use of new technologies; these new technologies will also enable the creation of robot bodies to house these fast computers. Such technologies include nanotechnology, atomic (quantum) computing, transmutation of elements, and artificial life.

Clearly, then, according to our authors, robots will be very intelligent. Given that such intelligent compact computers in robot bodies will be available within the first half of the twenty-first century, the next question is one of whether they will be able to be conscious persons. If a human plans on transferring a mind into such a robot, the robot must be able to support the mind and other aspects needed for being a person.

One view of the mind that is consistently rejected in the discussions of our authors is the view that the mind is a soul or substance distinct from the human brain. This view is seen as an outmoded religious or superstitious position. To our authors, the mind is seen more as the brain, various patterns in the brain, or the "program" running in the brain. The analogy here is with computer software. The mind's relation to the brain is like a program's relation to hardware running it. So in this respect robots should have no trouble having a mind, since the mind will be just the software running their robot brain hardware. What of consciousness and subjective experience? The view is that consciousness has a physical basis, so the claim is made either that such robots will be conscious or that we will come to believe they are conscious because they will act as if they are. Robots will also exhibit emotions. Though they don't put it in these terms, our authors clearly believe that robots will have all essential aspects of personhood.

So we have these intelligent, conscious robots available--how exactly do we get our minds into them? There are several stages to this, and for each a number of scenarios are presented. First, we have to find out what is in an individual's mind. Kurzweil notes that invasive (destructive) or noninvasive techniques might be used. Invasive procedures would include tearing apart the brain while observing its structure. Noninvasive techniques would include scanning, for example sophisticated MRI or optical scanning. Future MRI and optical scanning will be at the neuron or sub-neuron level and will allow observation of individual neurons firing. But there might be other ways. Moravec suggests several possibilities for computers attaching themselves to humans, plugging into electrical circuits in the brain, and learning brain details by observing the neural traffic.

The second stage of the transfer is to build the robot and transfer the mind to it. We have to either first build a robot brain, and then transfer the results of the scan into that existing robot brain or else do these two steps simultaneously. Most scenarios seem to follow the latter method; they depict not an already constructed robot awaiting a transfer but the creation of the robot brain equivalent to the human brain of the person transferring, and so the transfer seems to be the same process as the robot brain construction. But on a few scenarios we could break this second stage down into two distinct stages, with the second stage being the construction of the robot and the third stage being the transfer.

Our authors mention a lot of possibilities, and so under both schemes there may be multiple ways to accomplish the construction and transfer. Some depictions have the robot built to exactly copy the structure of the human brain of the person transferring, and when the robot is finished the mind has been effectively transferred. Other scenarios suggest the reverse engineering of the program or programs running the human brain. Once the program running on a particular person's brain has been determined, equivalent code can then be written for a robot brain. This robot brain might already be built, and then programmed with the equivalent code, or it might be built and programmed at the same time. In either situation, the structure of the robot brain would not necessarily mimic that of the human (whether at a fine level of detail ("fine-grained") or a coarse level), but the overall functioning would be equivalent. Furthermore, on another variation, if the scheme is to build the robot brain "on the fly" as the human brain information is being obtained, either the robot brain could be built physically separate from the human brain or parts of the robot brain built and substituted for parts of the human brain in a series of transplants.

In general the authors believe that personal identity will be preserved during the human-computer mind transfer. The consensus is that one's identity does not consist in having the same body or brain over time but in something else. It is a little difficult to say what that something else is, though. It is variously described as a continuity of "patterns of matter and energy," or of mental patterns, along with a continuity of memory. The basic idea is to preserve in the transfer whatever it is that gives each of us our sense of a self-conscious awareness of being the same person through time.

On the other hand, there are some signs of the realization that puzzles about identity might arise, though little attempt is made by our authors to address them, much less solve them. For example, if the transfer occurs by building a robot "brain equivalent" of my human brain through noninvasive scanning, then the robot could be fully functional while my human brain and body still are as well. The transfer has allegedly taken place by the time the robot was completed. So it seems we have two of me--the one with the robot body and the one with the human body. This might be seen as a difficulty, or at least a puzzle, concerning my identity--which one am I, possibly both? Our authors are aware of this possibility, and sometimes mention that we might experience puzzlement at it, but they do not provide a clear answer to the question raised. On the other hand, if my human body were destroyed some time before the robot were created, it is not clear whether the robot would be me or some other new person. So timing might be important to whether "I" go on or not. But how much time would have to elapse between the destruction of me and the creation of the robot before it would not be me but a new person? A split second? A day? Two weeks, etc.? This is another difficulty or puzzle that clearly arises on some transfer scenarios but no clear answers are provided.

Moravec seems to think that you can keep various copies of yourself stashed around the universe, so that if one of them is destroyed, you can still go on living as all the others. And if all are destroyed, other people or robots can pull out the blueprints and build you a new body and brain on the spot, with this body and brain really still the old you continuing to survive. Obviously in his discussion Moravec presupposes particular positions on a lot of controversial issues about personal identity. For example, it is certainly possible to wonder whether all these copies are really just "relatives" of some sort rather than you.

This completes our preliminary sketch of the extraordinary future. In coming chapters I turn to consider the view in more detail and subject it to a more thorough examination.


Copyright: 2000