NETFUTURE Technology and Human Responsibility -------------------------------------------------------------------------- Issue #148 August 5, 2003 -------------------------------------------------------------------------- A Publication of The Nature Institute Editor: Stephen L. Talbott (firstname.lastname@example.org) On the Web: http://www.netfuture.org/ You may redistribute this newsletter for noncommercial purposes. Can we take responsibility for technology, or must we sleepwalk in submission to its inevitabilities? NetFuture is a voice for responsibility. It depends on the generosity of those who support its goals. To make a contribution, click here. CONTENTS --------- Quotes and Provocations When the Mind Dogmatizes about Itself Intelligence and Its Artifacts (Stephen L. Talbott) Habits of the Technological Mind #1 DEPARTMENTS About this newsletter ========================================================================== QUOTES AND PROVOCATIONS When the Mind Dogmatizes about Itself ------------------------------------- Nearly every problem of a technological society is at bottom a problem of mind. In fact, this is true of society in general -- although the issues are often posed most sharply by technological developments. In future newsletters I will be dealing more and more, and in more fundamental ways, with the relation between mind and machine. So it may not be a bad time to make the following point, which seems to me both straightforward and simple. Anyone today who approaches the problem of mind by proclaiming himself committed to a materialist or mechanistic solution is a religious zealot. Typically, the proclamation runs something like this: "Science has explained the entire world based on physical principle alone. It makes no sense to appeal to some mysterious new principle in order to account for the last residue of unexplained phenomena -- the so-called 'mental' residue associated with the human brain". Leaving aside the fact that we cannot imagine any law whose reality lacks a conceptual (mind-like) component, so that "physical principle alone" is never actually physical principle alone; and leaving aside the historical truth that the "last", obstinate bits of reality troubling existing bodies of explanation have often signaled revolution (think of Planck and black-body radiation, Einstein and the speed of light) -- even apart from these considerations, there is reason to view the materialist commitment as a flight into religious zealotry. After all, the "last residue" is in fact the first, grounding principle, inasmuch as the entire history of scientific explanation has resulted from the activity of the "residue". Far better, then, to assume that what we discover about the mind will clarify the truth and error in this history than to let the history blindly determine what we will accept as the truth of the mind. When we hear someone declaring that the results of the mind's past, unselfenlightened activity should effectively prejudice our future understanding of the mind itself, then we can be sure that dogma is compromising the objectivity and openness so crucial to science. SLT Goto table of contents ========================================================================== INTELLIGENCE AND ITS ARTIFACTS Habits of the Technological Mind #1 Stephen L. Talbott (email@example.com) It has happened to all of us. We are quietly tending to our own business when a disturbance of the atmosphere -- its subtle approach unbetrayed by the visible trembling of even the most delicate leaf -- suddenly shakes us with preternatural force. Our reaction is unpredictable before the fact: anger, a shock of fear, laughter, surprise, a transport of joy -- all are possible. Often the atmospheric agitation may pass virtually unnoticed; but at rare moments we know instantly that our life will never be the same. Such is the cogent power imparted to the surrounding air when someone speaks a word. When you speak, you sculpt the stream of your breath, investing the sound-shapes as best you can with a measure of your own passion, determination, and insight. As with all worthy sculptures, these complexly structured sound-shapes carry within themselves some of the content of your inner life. They are in this sense bearers of consciousness. Even if I cannot see you, I may gain knowledge of you -- perhaps something about your innermost thoughts or feelings -- solely from the sculpted air-forms. I may then construct my own air-forms in reply, launching them into the atmosphere even before the reverberations of your speech have wholly died away. Earth's air mass is ceaselessly filled with these invisible sculptures, calling to each other, questioning, answering, provoking, sighing, mingling together and transforming one another in beautiful, infinitely complex patterns of interference. How differently we would think about them if they were routinely visible, as text is visible! If an alien were to visit our planet with appropriate instruments for detecting and decoding these patterns, presumably he would report back to headquarters: "Earth's atmosphere is a marvelous thing! It is filled with airy forms engaged in the most fascinating conversations. We must study the constitution of this unique air mass to determine how it evolved such intelligence". "Well", you say, "now you've gone too far. The bit about 'airy intelligence' is just plain silly". Yes, I went too far -- quite intentionally. Let's recognize the silliness. But let's also acknowledge the reason for it: in that last remark I failed to distinguish properly between the originating, human intelligence of the speaker, and the resulting, derivative intelligence impressed upon the air-forms. Yes, these forms really are bearers of intelligence -- a truth often underestimated. But at the same time they are derivative. It is one thing to speak and quite another to be spoken. Obvious, perhaps. Yet the failure to make this distinction between speaking and being spoken is taking root within our culture as a defining element in our stance toward intelligent machinery. Two Poles of Intelligence ------------------------- Audible speaking is not the only way we invest intelligence in the material of the world. We can write words upon paper, or print them using mechanical devices. Or we can encode them into magnetic tape, arrays of transistors, or the pitted surface of a compact disk. There are other ways we "speak" as well, impressing our wordless (but still word-like) intelligence upon the world. A screwdriver and plow bear the imprint of the functional idea motivating their invention. The same thing is true of vastly more complex machinery, where not only the static form of the parts but also the pattern of their movement and interaction is a bearer of intelligence, as in the printing press, loom, automobile engine, mechanical clock. And so, too, when the primary parts are minute stores of electrical energy and the movements are current flows, as in a computer. Now, until the past several decades we had little difficulty instinctively recognizing a twofold truth. First, it is perfectly natural to be pleased or incensed at the words we read on a page because they are indeed intelligent words; they mean something. But, second, this meaning is ultimately the author's, not (in the same sense) the paper and ink's. Our pleasure and anger are, in the end, responses to a person, not a sheet of paper. This, in fact, is why my opening picture of interacting atmospheric forms probably struck you as strange. It's not that my description was in any way untrue, but only that we habitually attend to the expressing intelligence rather than to the outer body of the expression. The sound-shapes, like the ink on the paper, drop out of consciousness because our primary interest lies with the speaker's interior world of thought, feeling, and intention. We do not ascribe this interior to the sound or ink in the same way we ascribe it to the speaker or writer. The case is similar when we consider complex, moving machinery, such as the press that produces the printed page: we can hardly assign authorship and its responsibilities to the press. After all, the same sort of relation obtains between the press and its builder as we see between handwritten words and the writer. There is, on the one side, intelligence embodied in material, and, on the other side, the person contriving the embodiment. No one would quarrel with this. But somewhere along the way from the printing press stamping out inked pages, to the programmed computer generating screen displays, many began to imagine that we were now dealing with a potential for native, originating intelligence in machinery. This conviction has thrived despite the obvious truth of the matter regarding "lesser" devices such as typewriters or calculators; it thrives despite the fact that we can impose the derived sort of intelligence upon matter without any apparent limit to the intricacy and complexity of these artifacts; and it thrives despite the inability of anyone to characterize in a principled way how such outward imprints of intelligence are transformed into an originating intelligence. Putting it in slightly different terms: we know from the simple and obvious cases that we need to distinguish intelligence from its artifacts. Both can be said to be intelligent, but not in the same way. How does this distinction between originating and artifactual intelligence disappear in the case of exceedingly complex artifacts? And why does the distinction, which must have some bearing on the complex devices whatever our view of them, receive so little attention? Certainly there is activity going on in our machines, just as there is activity going on between the intelligent sound-forms continually disturbing the atmosphere. But it doesn't seem too much to ask: how did this activity get there, and what is its relation to its source? A Fanciful Picture ------------------ The history of artificial intelligence has, for the most part, been a history of play with, and theorizing about, the outward bodies (or "tokens") of symbols. There has been widespread disregard of the inner act by which alone the token is constituted a symbol. And even when attention is given to the problem, the real issue typically remains ignored. In his textbook, Artificial Intelligence: The Very Idea, philosopher John Haugeland acknowledges what he calls "the mystery of original meaning", which he describes as the problem of "telling minds from books". But he frames the mystery this way: "suppose some special brain symbols had original meaning and all other meaningfulness derived therefrom". So he starts with the already materialized symbol and then asks whether it may, in some cases, have a special, original significance. It is exactly as if he were to inquire whether some of the words on the page or some of the spoken sound-shapes have a special, original significance. This is to look in the wrong place. It is to look for the speaker in his speech. Needless to say, Haugeland finds no evidence for his "original" meaning. But neither can he disprove its existence, so he finally resorts to a bit of science fiction. Suppose, he writes, that a future comes when intelligent computers are ensconced in mobile and versatile bodies; and they are capable (to all appearances anyway) of the full range of "human" communication, problem solving, artistry, heroism, and what have you. Just to make it vivid, imagine further that the human race has long since died out and that the Earth is populated instead by billions of these computer- robots. They build cities, conduct scientific research, fight legal battles, write volumes, and, yes, a few odd ones live in ivory towers and wonder how their "minds" differ from books -- or so it seems. One could, I suppose, cling harshly to the view that, in principle, these systems are no different from calculators; that, in the absence of people, their tokens, their treatises and songs, mean exactly nothing. But that just seems perverse. If [artificially intelligent] systems can be developed to such an extent, then, by all means, they can have original meaning. But, as Haugeland is doubtless aware, this is no argument. It is simply a way of gratifying himself by picturing vividly what he would have liked to prove. When he imagines that artificially intelligent systems "can be developed to such an extent", he ignores the fact that the developing intelligence must stand upon a different level from what is developed and, crucially, the difference has to do with a power of acting versus the products of this power. What makes Haugeland's picture compelling for many people, I think, is their vivid awareness of technical advance, their current experience of machines that "talk" and do other human-like things, and their inability to imagine any clear limit to how intricate and sophisticated our devices may become. That last is a vital point. There is no clear limit to how intricate and sophisticated our devices can become. It is one of the glories of humanly intelligent creatures that we can organize matter in increasingly complex ways. All that is needed, then, is for us to forget ourselves -- to ignore our activity at its originating source -- and then to project onto our machines what has thus been lost to our own consciousness, whereupon we will find Haugeland's fictional scenario absolutely convincing. We will also find ourselves without any basis for comprehending what robots might and might not actually do -- expectations that depend upon our ability to distinguish thinking from its embodiments. This ignoring of ourselves and projection of our highest activities upon machines are already entrenched features of our culture. Mechanization of the Word ------------------------- Every word presupposes a speaking consciousness. The word is a bearer of consciousness; there is always an expressing agent behind it. This is why we feel instinctively compelled to respond to computers as if we were speaking to persons, just as we respond to a book as if we were speaking to the author. The instinct is a healthy one -- we really are responding to someone's speech -- but the health of the instinct depends on our ability to identify correctly who the speaker is. The problem is that, compared with face-to-face conversation, the book had already introduced a great distance between reader and author. And now, between the programmer specifying words to be assembled upon patterns of logic, on one side, and the user interacting with computer output on the other, the anonymous distance has grown hopelessly large and indeterminate. This distance has, moreover, been rendered opaque by the abstruse complexities of digital technology. No wonder many abort their effort to understand the source of the words and instead ascribe the machine's obvious intelligence to the machine itself as agent. You can see the situation we're getting into. From the supermarket checkout counter to the bank teller to web transactions to the home entertainment center, our lives are ever more thoroughly embedded within automated, intelligent systems. We are continually assaulted by the mechanized word, but less and less aware of where the speaking comes from. We have less and less experience of the conscious acts giving rise to the words, and we ourselves are not really addressed by the words. It is "the System" that speaks, but the System seems to be nothing more than the machine rendered vague and ubiquitous. Given such an intense training to disregard the originating speaker, is it any wonder that our own speaking, as non-mechanical act giving rise to the embodied symbol, has been disappearing from our awareness? Degradation of the Spoken Word ------------------------------ In some ways we have not overestimated intelligence-as-product, but rather underestimated it. The intelligence we invest in our artifacts is really there -- out there in the world. This is a truth that, in good Cartesian fashion, we have spent several centuries trying to ignore. We wanted a world of inert things, absolutely devoid of any interior -- pure outward matter. And so nature was disensouled, becoming the domain of things rather than speech and expression. We progressively lost our sense for the living qualities of our environment. Our attention turned from the inner meaning of things to their outward, syntactic forms -- rather like analyzing the patterns of ink on a sheet of paper instead of reading the words. Having forgotten the speaking, we inevitably began to lose a proper appreciation for what was spoken as well. Nature died for us. This, I suspect, is why the latter-day invention of intelligent machinery has struck us with such galvanizing force. Even as a caricature -- even if we have mistaken the spoken machine for our own powers of speech -- our attention is being redirected to the word-like character of the world. There is a kind of re-ensouling going on. Or could be. A caricature can be a step toward truer understanding, or else a step toward an ever more radical disfiguration of the truth. As good a way as any to grasp the divergent potentials of our current situation is to look at what has happened to human speech. When the spoken word was alphabetically encoded and written down, it became a much thinner bearer of intelligence, since it almost completely lacked the expressive potentials of the voice's artistic sculpturing. The relation between the outward "body" of the sign and its inner meaning shifted toward arbitrariness. The sculptural and artistic aspect, whereby the outward body is itself an expression of the inner meaning -- the meaning made perceptible -- is no longer there. Words, all the more as handwriting becomes machine printing, perform the barest pointing function: "This sign stands for that meaning". There is no particular reason for the sign to stand for the meaning, because the sign is no longer the meaning condensed into external form. In this context, intelligent artifacts appear less as an occasion for re- ensouling the world than as a symptom of the dis-ensouling even of our own speaking. An alien would be much more justified in saying the earth's sounding air mass is intelligent than in saying a computer is intelligent. The sound-forms are more fully expressive of the entire human being than is digital logic. There are various logics implicit in the forms, but the forms are much more than the logic. I have personally experienced the expressive degeneration of language by listening to the National Weather Service radio broadcasts ever since, a year or two ago, they began to feature computer-synthesized voices. Sometimes I listen several times a day. It suddenly hit me recently that I had almost completely acclimatized to the artificial voice. True, when I actually stop and attend to the voice, I realize there hasn't been a great deal of improvement since the automated broadcasts began. There are the same mispronunciations, the same punctuational and inflectional absurdities. But the effort of attention required in order to notice these distortions becomes ever more difficult. I am being habituated to a degraded language. But this raises an interesting question. What might the opposite process -- a genuine re-ensouling of language -- look like? If in fact we are the ones who speak technology into existence, how could our speaking become more profound? Facing Up to Disability ----------------------- We are all continually subjected to uninspired, lifeless speech, as well as to the artificial inspiration lying behind calculated rhetorical effect. But we have also noticed on occasion (even if the noticing was scarcely conscious) what a remarkable difference a genuinely talented actor or storyteller can make. Try listening for example to Nicol Williamson's reading of The Hobbit. In such cases the words come alive and the listener may be transfixed by them. I believe we have paid entirely too little attention to the objective differences of speech that reliably produce such different listening experiences. Unfortunately, in an age preoccupied by syntactic structures and information, these qualitative differences -- which are always substantive differences in content and meaning -- scarcely rate much serious attention outside performance circles. Certainly they do not often enter into discussions of artificial intelligence. But this is exactly the problem. I would like to illustrate it by relating an experience of my own. A few years ago I had a brief opportunity to work with Patricia Smith, a speech artist who had spent years in a special speech training. Her aim was to recover those depths of language where, to some degree at least, the sounds of words are themselves the expressive embodiment of the words' meanings. These depths are mostly forgotten, and therefore are commonly denied by linguists. But this is less a point for arguing than for experiencing. My own experience was startling. To hear Patricia recite anything -- a child's poem, say -- was to hear a power issuing from interior regions I had been completely unaware of, a power clothing itself in sound. It's not that she recited with great outward force or drama, in the usual sense. The quiet power was that of the words themselves expressing some of the potential with which they were invested at their forging. Actually, most people would notice nothing very special upon hearing Patricia speak -- just as, in the other direction, I typically hear nothing unusually degenerate in the weather broadcasts. Our hearing is dulled; we mostly care only for the abstract content we call "information". It was not until I tried to repeat lines after Patricia that I was forced to listen and hear. Even then, it took a very long time for me to recognize and accept the disturbing reality I was being presented with. You might think that anyone, with a little work, could learn to repeat at least an isolated line with roughly the same expression as another person. It never occurred to me that the case might be otherwise. Yet after several months I had to acknowledge that I could not come near what I was slowly learning to hear in Patricia's speech. No matter how long I worked on one sentence, one phrase, a single word, I still heard this heart- sinking deadness in my own voice compared to hers. I simply could not imitate the simplest sound in any of its profundity. For someone who had come to think about sound, like nearly everything else, in a much too mechanical way, this truth was not easily accepted. "Just shape your damned vocal chords the right way, and you'll get the quality of the sound you want". But no, I had to realize that to sculpt a fully expressive word would be no less an achievement of imagination, beauty, and meaning than Picasso's sculpting of metal or Mozart's crafting of a sonata. I had no realistic prospects of getting very far. Because of my own peculiar and severe biographical limitations, I am in many ways an extreme case. But in general we do not prepare our young for the realization of their speaking potential. Rather, we do everything we can to destroy it. If you doubt this, just listen the next time your kid is playing with a "talking" doll or a video game. Or listen to the weather radio. Learning to Speak Again ----------------------- We know, or could know, that a sick patient residing in a beautiful, restful setting, feeling the sun and breeze, will do better than in a typical hospital setting. We know that the sound of the human voice in song -- now often employed in hospices -- can be dramatically healing. We know that different colors can quite literally move us in different ways. We know that architecture can be drab and oppressive (look in most schools), or else can encourage the human spirit to soar. In these and countless other ways our speaking -- the qualities of it, the inner expressive power we manage to invest in the stuff of the world -- is a force for good or ill in society. Really, developing this power of creative speech is the only thing our life is about. And it goes hand in hand with our ability to hear the world's speech. If we were to attend to the full expressiveness of the human being with even a fraction of the resources we devote to extracting dessicated "information" from the genome, might we not gain far greater powers for health than we will ever gain from mechanistically conceived genetic manipulations? A healthy technology will be as much art as machinery. Currently, however, the elimination of the qualitative and artistic dimension from consideration, leaving us with a naked skeleton of logic and algorithm, is more or less what defines technology. The voice proper, I think, gives us the best model of our technical and creative potentials. There is a kind of living power in the spoken word greatly transcending the merely technical, which is why it can so directly move us and light up within us as an awareness of another person's interior. And it is not only that the sounding voice moves us; it also has the power to shape the stuff of the world, calling it up into complexly organized artistic forms. (If you're interested in this shaping by sound and related vibrations, look into the "cymatic" researches of Hans Jenny and the literature on Chladni figures.) The voice can both shatter crystal and impose qualitative form upon a delicate flame. Who knows how deeply our creative impulses might penetrate the mysterious interior of matter, merging with the most primal speech underlying things themselves, if we should begin cultivating an awareness of the full range of qualities in our own voices. We will have to overcome several centuries of strict training if we are to explore these new possibilities. Many of us will have to say, as I had to say about Patricia's speech, "This is mostly beyond me; as a child of the technological age, I have been closed off from too many of the prerequisites for getting very far with such a skill". And no doubt some of those unwilling to accept this truth will become snake oil salesmen, purveying endless nonsense under cover of the potentially vital disciplines of aromatherapy, chromotherapy, feng shui, and all the rest -- just as, on the other side, we have heard endless nonsense about curing diseases from the genetic engineers. I have no doubt that there are a handful of wise practitioners in the various disciplines just mentioned. And I do not intend my own experience with speech to become a counsel of hopelessness. I am a worse case than most! And we can always take profitable steps in the right direction. Personally, I count it a great gift to have heard Patricia Smith's recitation, and to have gained my first inkling of a world of expression, a world of speaking, previously lost to my hearing. I am convinced that it is vitally important for our society as a whole to gain some such inkling. Otherwise, there will be nothing to restrain the death rattle, the hollow crescendo of empty symbols in apparent conversation with each other, but with no one speaking. A Concluding Note to AI Researchers ----------------------------------- Here is a thought experiment: Suppose we have two robots inside a room, conversing. You and I are able to observe the room, but in different ways. One of us directly watches the robots and hears their dialogue; the other is not allowed to see them but is given technical means to observe the air-forms in the room and to hear these forms as speech. The task of the one observer is to decide whether the robots are intelligent in anything like a human sense, while the other observer must decide whether the air is intelligent. One of the various questions we can ask, then, is whether the prevailing notions of robotic intelligence allow us to call the robots "intelligent" in a way that does not apply to the air. I know full well that a thousand objections to such a "preposterous" thought experiment will immediately leap to your mind. But I would urge you to stick with it for a while and try to think the matter through a little further. For many of the objections will not prove as straightforward as they appear at the outset. I can, for example, imagine someone replying: Of course there's no intelligence in the air as such. It's artificial to treat the air in isolation from the physical structure of the robots, since there is a clear, consistent, cause-and-effect relation between the vocal actions of the robots and the movements of the air. It's always an overall, coherent, and interrelated system that is intelligent. Fine. But if then we seriously want to consider the atmosphere as part of our intelligent, robotic system, a great deal seems to follow. Is there any way in principle to isolate the electrical currents of the robot's "nervous system" from the regional generating station and power grid? Must we then consider the power station (or the sun, if solar panels are used) as an essential organ of our intelligent robot? And, even more obviously, there are the programmers, hardware engineers, and others whose actions are in various regards causally responsible for building and sustaining the robot and specifying its patterns of action. Within the discipline of artificial intelligence there has been great disdain for the popular sentiment that says a computer "can only do what it's programmed to do". But if, as the researchers would prefer, we disregard the programmer as a crucial part of our intelligent robotic system, why should we not disregard the robot as part of an intelligent atmospheric system? Think of it this way: it's perfectly reasonable to try to narrow down your definition of an intelligent system. The question is whether the body of the robot qualifies for such a narrowing down any more than the body of air. The fact is that the robot is mortised into a causal nexus that extends outward to include just about everything. So long as we concern ourselves only with the products of intelligence (leaving originating agents out of account) we have no clear ground for locating a source for any particular intelligent act. The entire world becomes for us a kind of robotic device. This is, in fact, the direction in which I see us being driven. It is no accident that even materialists have taken up the language of intelligence -- and particularly the terminology of information -- and begun to apply it with little sign of embarrassment to the entire natural world, whether they are speaking of evolution or black holes. There is truth in this, or would be if the essential effort were not to deny active intelligence while taking covert advantage of its explanatory value. There would be truth in it if we were able to distinguish between consciousness and that of which it is conscious, between the act of thinking and its products. But the main thrust of AI today is to reconceive consciousness and thinking in terms of their results, forgetting our own activity as conscious thinkers. I spoke above of the "originating agent", and this will be the point at which many become puzzled today. What is meant by such an agent, or by thinking as act rather than thinking as product? Suffice it for the moment for me to point out that in the course of evolutionary history, long before anyone was staking out personal positions or thinking about human-machine relationships, the members of one particular species began, from somewhere within themselves and without philosophical prompting, to say "I". Surely one place to look for an originating agent is in the speaker of this word, and in the interior activity lying behind the word. And certainly the only place we can discover the essence of this activity is in the self that performs it -- that is, in the "I", not in the material symbols every "I" proceeds to fashion from the world. As for our thought experiment, the "intelligent-atmosphere" hypothesis seems to me at least as defensible as the "intelligent-robot" hypothesis -- there really is intelligence in all our artifacts. But the two hypotheses also suffer the same limitation: it makes no more sense to say the robot is an originating intelligence than to say the atmosphere is. Related articles: ----------------- "And the Word Became Mechanical", in The Future Does Not Compute: http://netfuture.org/fdnc/ch18.html "Mindlessness and the Brain", in NF #138: http://www.netfuture.org/2002/Nov0702_138.html Goto table of contents ========================================================================== ABOUT THIS NEWSLETTER Copyright 2003 by The Nature Institute. You may redistribute this newsletter for noncommercial purposes. You may also redistribute individual articles in their entirety, provided the NetFuture url and this paragraph are attached. NetFuture is supported by freely given reader contributions, and could not survive without them. For details and special offers, see http://netfuture.org/support.html . Current and past issues of NetFuture are available on the Web: http://netfuture.org/ To subscribe or unsubscribe to NetFuture: http://netfuture.org/subscribe.html. Steve Talbott :: NetFuture #148 :: August 5, 2003 Goto table of contents
This issue of NetFuture: http://www.netfuture.org/2003/Aug0503_148.html.