4. Intellectronics
Return to Earth
The megabit bomb
The big game
Myths of science
The intelligence amplifier
The general trend to mathematize the sciences, including those which up to now did not use mathematical tools traditionally, after biology, psychology and medicine slowly includes even the humanities, although for the time being only in the form of rather isolated guerrilla actions which can be observed, e.g., in the field of (theoretical) linguistics or literary theory (application of information theory to the investigation of literary texts, especially - poetry). At the same time, however, we meet the first signs of an unusual and rather unexpected phenomenon, namely the fact that mathematics (of every kind) turns out to be insufficient for the realization of certain, only recently formulated goals at the frontier of the most advanced of all the latest research activities: we are talking about the problems which self-organizing homeostatic systems are facing. Let us mention, rather exemplarily, some basic problems in which specialists first met those shortcomings of mathematics. These are - the construction of an intelligence amplifier, a self-programming control device for the industry, finally - this is the most general task - a universal homeostate of a complexity which is comparable to our own, human complexity.
The intelligence amplifier, probably proposed for the first time as a real construction program by Ashby [W. Ross Ashby: Introduction to Cybernetics (?)], is supposed to represent the exact equivalent to the amplifier of physical strength, which is every human-controlled machine, in the area of mental activities. Amplifiers of strength are cars, excavators, cranes, machine tools, generally every device in which man "is connected" to the control system as a source of control, but not of strength. Contrary to all impressions the deviations of individual intelligence levels from the average are not larger than those in the field of physical capabilities. The average I.Q. (measured mostly by application of psychological tests) ranges from about 100 to 110; highly intelligent persons reach 140-150, and the upper, extremely seldom found limit is about 180-190. Hence an intelligence amplifier of more or less the same factor as the average multiplication of an industrial worker's strength by the machine which he operates, would have an I.Q. of about 10000. The possibility of constructing such an amplifier is no less real than the possibility of constructing a machine which is a hundred times stronger than a human. Of course the chances of its construction are not very large for the time being, for the most part because of prime importance rather is the building of another device - the already mentioned control device for the industry (a "homeostatic brain for automated factories"). However, I will dwell on the example of the amplifier because it makes it easier to demonstrate the fundamental problem with which the engineer is confronted here. The thing is that he must construct a device which is "more intelligent than he himself". It is obvious that he would not solve the posed problem if he wanted to tackle it by proceeding according to the traditional method of cybernetics, i.e., making an appropriate action profile of the machine, because this profile already describes the limits to the "intelligence" that the constructed device could reach. Apparently - but only apparently - the problem seems to represent an insoluble paradox, the kind of proposition to lift oneself up by one's own hair (and, additionally, having a hundred-ton weight tied to one's feet...). Indeed, the problem is insoluble, at least according to today's criteria, if we were to postulate the necessity of making an inevitably mathematical theory prior to the construction of the amplifier. There is, however, a wholly different approach to the task, for the time being known only as a hypothetical possibility. The specific knowledge of the internal design of the intelligence amplifier is not accessible for us. Maybe it is totally dispensable. Maybe it is sufficient to treat that amplifier as a "black box", as a device about whose internal plan and successive states we do not have the slightest idea, since we will only be interested exclusively in the eventual results of its action. That amplifier, like every cybernetic device taking pride in itself, possesses "input" and "output". Between them extends the area of our ignorance, but what the heck, if only this machine really behaves like an intellect with an I.Q. of the order of 10000?
Since the method is new and, up to now, nowhere applied, it sounds, admittedly, a little like a concept from an absurd comedy rather than something like a technological production recipe. But here are some examples which might make its application more probable. One can, let us assume (this has been done), give some iron powder to an aquarium in which there is a colony of infusoria. The infusoria absorb some of the iron together with their food. Now if we apply an external magnetic field to the aquarium it will in a certain way influence the motion of the infusoria. Hence the changes in field strength represent "signals" in the "input" of our "homeostate", the states of the "output", however, are determined by the very behavior of the infusoria. It is not important that, for the time being, we do not know what that "infusory-magnetic" homeostate could be used for, nor that in this form it has nothing in common with the hypothetical intelligence amplifier. The important thing is that although we do not know anything about the real complexity of a particular infusorian, although we are not able at all to sketch its design, the way one sketches a machine design, we still succeeded to build from these elements, of which we had no detailed knowledge, a certain higher-level whole, subject to system laws, possessing signal "input" and "output". In place of the infusoria one could use, e.g., certain types of colloids or send an electric current through a multiphase solution, as a result of which certain substances might be precipitated, thus changing the conductivity of the solution as a whole, which in turn might cause an effect of "positive feedback", i.e., amplification of the signal. We immediately have to admit that up to now these attempts did not give any revolutionary results, and there are many cyberneticians who disapprovingly watch this heretical deviation from the traditional treatment of electronic elements, this search for new substances, new building materials which in certain aspects are similar to those of living organisms (which is by no means coincidental!). [G. Pask: A Proposed Evolutionary Model. In: Principles of Self-Organization. Transact. Univ. Illinois, Symposium on Self-Org., Pergamon Press, 1962]
Without anticipating the results of such investigations now we understand a little better how one can, from "incomprehensible" elements, build systems which work as desired. Here, at the very basis of engineering activity, a fundamental methodical shift is taking place. Present engineering behaves a little like somebody who is not even trying to jump over a ditch as long as he did not theoretically establish all relevant parameters and their dependencies - i.e., as long as he did not measure the strength of local gravity, the performance of his muscles, as long as he does not know the exact kinematics of the motions of his own body, the characteristics of the control processes in the brain, etc. The technologist-heretic of the cybernetic school, on the other hand, simply tries to jump over the ditch and assumes, justifiably, that in case he succeeds, the problem has been solved properly.
In doing so he refers to the following fact. Any physical activity, like the above mentioned jump, requires preparatory and executive brain work, which is nothing else than only an immensely complicated sequence of mathematical processes (to which in general any activity of the brain's neural network can be reduced). However, the same jumper, "having in mind" all that cerebral mathematics of the jump, will generally not be able to write down on paper its theoretical-mathematical counterpart, which would consist of the appropriate number of exact formulas and transformations. This seems to follow from the fact that this "biomathematics" which is applied in general by all living organisms including the infusorian, for its mathematical verbalization in a classical, school or university sense, requires the repeated translation of whole systems of impulses from one language into another - from the nonverbal and "automatic" language of biochemical processes and the flow of neural excitations into a symbolic language, whose formalization and construction are the task of totally different parts of the brain than those which directly control and realize that "innate mathematics". Hence the key to the problem actually is that an intelligence amplifier would not have to formalize, construct, verbalize, but that it worked automatically and "naively", but at the same time as reliably and infallibly as the neural processes of our jumper - that it did nothing but transforming the stimuli entering through the "input" in order to deliver the complete solution through the "output". Neither itself, that amplifier, nor its designer, nor anybody else will know how it is doing that - but we will obtain what we exclusively require: results.
The black box
On the ethics of homeostates
The dangers of electrocracy
Cybernetics and sociology
Belief and information
Experimental metaphysics
Beliefs of electrobrains
The spirit in the machine
Problems with information
Doubts and antinomies
...11/23/97, to be continued..