With Transcendence opening in movie theaters today, I thought it was a good time to continue here with my posting of science fiction reviews I first published in the 1990s, this time of Charles Platt's The Silicon Man, first published in 1991, brought out in paperback in 1993, and reviewed by me that year in a highly philosophic analysis in the Journal of Social and Evolutionary Systems.
In Mind at Large: Knowing in the Technological Age (1988, p. 180), I wrote that "A flat denial forever and anon of AI possibilities [humanlike intelligence] in nonliving circuits amounts to ... an unbecoming protein chauvinism." What I had in mind was the dogmatism of a position that says, just because the only intelligence we know is protein-based, therefore all intelligence must be so. I went on, however, to depict the attempt at creating real intelligence in computer systems as akin to putting "Descartes before the horse," by which I meant that since intelligence is a property of life insofar as we know, we're more likely to develop artificial intelligences out of artificial living entities than from any non-living artificial components.
Charles Platt's The Silicon Man, a science fiction
novel, takes up the challenge of AI from another angle.
Pointing out that something (like the mind) can be copied
without the copier fully understanding how the original entity
works ("You mean Gottbaum and his people copied my brain without
knowing how some of it works?" "Yes. By analogy, an audio
recorder can copy a piece of music without understanding harmony
and composition. All that matters is that the copy is
accurate," p. 147), Platt explores the implications of uploading
a person's mind into a central computer. Of course, the analogy
is imperfect -- music, regardless of its complexity and unlike
the mind, is not a self-regulating, generative system -- and our
scientific capacity to do this is vastly beyond our current
grasp. But the lack of ipso facto impossibility of Platt's
scheme -- an impossibility that one could take refuge in only on
the basis of a protein chauvinism -- makes it and the book it is
in worthy of very serious philosophic contemplation.
The central philosophic issue it raises for me is, given that a human intelligence could be copied into a computer whose system could supply that intelligence with one hundred percent accurate simulations of everything ranging from making love to fine dining to evening breezes, what differences if any would be worth claiming between this simulated existence and its original "real" one? A related ethical issue is, given that such differences are negligible, would termination of fleshly existence in favor of silicon constitute murder, if involuntary, or otherwise suicide?
Platt's book focuses more on the ethical issue, raising the stakes by suggesting that a human intelligence in a computer might even be an existence superior to the old-fashioned one (for example, "infomorphs," intelligences in a computer, don't age, p. 223).
But I find the ontological question more primary. In several essays (1994a, 1994b, 1994c), I've delved into questions of what can't be done in cyberspace, though from the perspective of a flesh-and-blood body working in and through cyberspace (as anyone connected to any computer network can now do), rather than the intelligence in the body literally vacating it in favor of a total intra-cyberspace existence. My point in these essays is that in areas in which the body must be served -- as in making love, leading to procreation, and eating for nutrition -- then whatever completely convincing alternative cyberspace can provide is obviously not enough.
But what about the human intelligence totally within the computer? Can it be fully served by its simulations, and if so, what does this say about the relationship of human intelligence and the external material universe from which it emerged? Here we come upon the most fundamental questions about the nature of reality and its relation to perception. We can start with the tedious observation that, yes, we have no idea that what we perceive in our current external world is really there -- the whole universe could be our dream -- but once we move beyond this logically irrefutable but fruitless observation we're left with a very profound distinction between reality perception and simulated perception. The first is a relationship of perception (and the perceiver) to something not of its own making (well recognized by Kant's insistence that knowledge is a product both of our internal cognitive processors and the external data they work upon); the second smacks of Narcissus looking at endless mirrored reflections of his own mind. And thus the second kind of perception -- the perception of human intelligence wholly internalized in cyberspace -- seems to return to the sterile solipsism of the world is my dream.
Platt is aware of this issue, having one of his computer-internalized characters observe that "from the inside, as infomorphs, we obviously can't alter the structure -- the actual hardware -- of [our central computer]. That would be like a tape recording trying to alter the structure of the tape on which it was recorded" (p. 236). Actually, the analogy isn't the best, since a tape recording with a very loud sound might in principle cause a speaker to blow, which could in turn cause the tape-turning mechanism to malfunction, which might in turn alter the tape -- but it nonetheless suggests that even a perceptually intra-cyberspace existence, totally inside cyberspace, requires the existence of outside, real "hands-on" ministration (if, say, the hardware of the central system is in need of repair).
Near the end of the novel, though, Platt imagines the
growth of infomorphs in computer networks achieving such power
that they can in fact control physical events outside of their
systems. ("You can rent [a vehicle], pipe your mind into it,
and go wherever you want if you still need to interact with the
real world," p. 255.) And this confronts us with what might be
the most fundamental ontological question of all: If we can
indeed copy everything -- every aspect of an entity -- then is the
copy in any sense a copy, or is it better thought of as another
original?
Well, there is what I call in Mind at Large (pp. 149-150) the paradox of copying: the copy, to the degree that it is a perfect copy, defeats itself because in so being a perfect copy it transforms the original into a duplicate, and therein the perfect copy is no longer a perfect copy (because it has obliterated rather than preserved the uniqueness of the original, and therein failed to copy a central aspect of the original). A perfect, artificially constructed human intelligence would inevitably have this effect on its natural progenitors.
On the other hand, there seems room enough -- and need enough -- for both of us in this universe. I recommend Platt's book for stirring attention to such issues. Like Isaac Asimov's robot series, it shows that, in constructing our future, we need not only technology and philosophy but its presentation in science fiction.
References
Levinson, P. (1988) Mind at Large: Knowing in the Technological Age. Greenwich, CT: JAI Press.
Levinson, P. (1994a) "Will the Delta Clipper Turn Deep Space Into Cyberspace?" Wired, February, p. 68
Levinson, P. (1994b) "Picking Ripe: There Are Just Some Things You Can't Do In Cyberspace." Omni, August, p. 4.
Levinson, P. (1994c) "Entering Cyberspace: What To Embrace, What To Watch Out For." Journal of Social and Evolutionary Systems, 17 (2), pp. 119-126.
In Mind at Large: Knowing in the Technological Age (1988, p. 180), I wrote that "A flat denial forever and anon of AI possibilities [humanlike intelligence] in nonliving circuits amounts to ... an unbecoming protein chauvinism." What I had in mind was the dogmatism of a position that says, just because the only intelligence we know is protein-based, therefore all intelligence must be so. I went on, however, to depict the attempt at creating real intelligence in computer systems as akin to putting "Descartes before the horse," by which I meant that since intelligence is a property of life insofar as we know, we're more likely to develop artificial intelligences out of artificial living entities than from any non-living artificial components.
Charles Platt's The Silicon Man, a science fiction
novel, takes up the challenge of AI from another angle.
Pointing out that something (like the mind) can be copied
without the copier fully understanding how the original entity
works ("You mean Gottbaum and his people copied my brain without
knowing how some of it works?" "Yes. By analogy, an audio
recorder can copy a piece of music without understanding harmony
and composition. All that matters is that the copy is
accurate," p. 147), Platt explores the implications of uploading
a person's mind into a central computer. Of course, the analogy
is imperfect -- music, regardless of its complexity and unlike
the mind, is not a self-regulating, generative system -- and our
scientific capacity to do this is vastly beyond our current
grasp. But the lack of ipso facto impossibility of Platt's
scheme -- an impossibility that one could take refuge in only on
the basis of a protein chauvinism -- makes it and the book it is
in worthy of very serious philosophic contemplation.The central philosophic issue it raises for me is, given that a human intelligence could be copied into a computer whose system could supply that intelligence with one hundred percent accurate simulations of everything ranging from making love to fine dining to evening breezes, what differences if any would be worth claiming between this simulated existence and its original "real" one? A related ethical issue is, given that such differences are negligible, would termination of fleshly existence in favor of silicon constitute murder, if involuntary, or otherwise suicide?
Platt's book focuses more on the ethical issue, raising the stakes by suggesting that a human intelligence in a computer might even be an existence superior to the old-fashioned one (for example, "infomorphs," intelligences in a computer, don't age, p. 223).
But I find the ontological question more primary. In several essays (1994a, 1994b, 1994c), I've delved into questions of what can't be done in cyberspace, though from the perspective of a flesh-and-blood body working in and through cyberspace (as anyone connected to any computer network can now do), rather than the intelligence in the body literally vacating it in favor of a total intra-cyberspace existence. My point in these essays is that in areas in which the body must be served -- as in making love, leading to procreation, and eating for nutrition -- then whatever completely convincing alternative cyberspace can provide is obviously not enough.
But what about the human intelligence totally within the computer? Can it be fully served by its simulations, and if so, what does this say about the relationship of human intelligence and the external material universe from which it emerged? Here we come upon the most fundamental questions about the nature of reality and its relation to perception. We can start with the tedious observation that, yes, we have no idea that what we perceive in our current external world is really there -- the whole universe could be our dream -- but once we move beyond this logically irrefutable but fruitless observation we're left with a very profound distinction between reality perception and simulated perception. The first is a relationship of perception (and the perceiver) to something not of its own making (well recognized by Kant's insistence that knowledge is a product both of our internal cognitive processors and the external data they work upon); the second smacks of Narcissus looking at endless mirrored reflections of his own mind. And thus the second kind of perception -- the perception of human intelligence wholly internalized in cyberspace -- seems to return to the sterile solipsism of the world is my dream.
Platt is aware of this issue, having one of his computer-internalized characters observe that "from the inside, as infomorphs, we obviously can't alter the structure -- the actual hardware -- of [our central computer]. That would be like a tape recording trying to alter the structure of the tape on which it was recorded" (p. 236). Actually, the analogy isn't the best, since a tape recording with a very loud sound might in principle cause a speaker to blow, which could in turn cause the tape-turning mechanism to malfunction, which might in turn alter the tape -- but it nonetheless suggests that even a perceptually intra-cyberspace existence, totally inside cyberspace, requires the existence of outside, real "hands-on" ministration (if, say, the hardware of the central system is in need of repair).
Near the end of the novel, though, Platt imagines the
growth of infomorphs in computer networks achieving such power
that they can in fact control physical events outside of their
systems. ("You can rent [a vehicle], pipe your mind into it,
and go wherever you want if you still need to interact with the
real world," p. 255.) And this confronts us with what might be
the most fundamental ontological question of all: If we can
indeed copy everything -- every aspect of an entity -- then is the
copy in any sense a copy, or is it better thought of as another
original?Well, there is what I call in Mind at Large (pp. 149-150) the paradox of copying: the copy, to the degree that it is a perfect copy, defeats itself because in so being a perfect copy it transforms the original into a duplicate, and therein the perfect copy is no longer a perfect copy (because it has obliterated rather than preserved the uniqueness of the original, and therein failed to copy a central aspect of the original). A perfect, artificially constructed human intelligence would inevitably have this effect on its natural progenitors.
On the other hand, there seems room enough -- and need enough -- for both of us in this universe. I recommend Platt's book for stirring attention to such issues. Like Isaac Asimov's robot series, it shows that, in constructing our future, we need not only technology and philosophy but its presentation in science fiction.
References
Levinson, P. (1988) Mind at Large: Knowing in the Technological Age. Greenwich, CT: JAI Press.
Levinson, P. (1994a) "Will the Delta Clipper Turn Deep Space Into Cyberspace?" Wired, February, p. 68
Levinson, P. (1994b) "Picking Ripe: There Are Just Some Things You Can't Do In Cyberspace." Omni, August, p. 4.
Levinson, P. (1994c) "Entering Cyberspace: What To Embrace, What To Watch Out For." Journal of Social and Evolutionary Systems, 17 (2), pp. 119-126.
No comments:
Post a Comment