|Intelligent Systems And Their Societies||Walter Fritz|
This letter, I fear, is long overdue. I intended to write when the discussion about the definition of science fiction was going on, and just never got around to it. Here is my definition: "A story placed into a time where science is, in some respect at least, farther developed than now".
Here is my reaction to the article "Thinking in man and machines". My definition of thinking is: "a process changing the future to suit the owner of the thinking device".
This definition is unusual and needs explanation. Thinking in man consists of using the external senses as one input, the bodily needs as the second input; then finding out from the external senses what the future will be, and moving the limbs in such a manner that the future will be advantageous to the body. (p.Ex. senses see food, the body indicates hunger, i.e. need for food). The useful future would be movement of the food into the body. Limbs are activated to stretch out, pick up, et cetera, so that the future - which without interference would have been an unchanged position of food and body - is changed into the useful future of food intake. (About the objection that man does not know the future: Every natural law is knowledge of the future, and that is the only usefulness of natural laws.)
About consciousness: lt seems to me that man is in a process of change for the last thousand years or so, and for quite some time to come. Let me explain. Once the application of a set of rules is figured out, a pattern called behavior pattern or habit is set up, and our unconscious sees to the proper actuation of the limbs. (p.Ex. the correct walking down the street we walk every day, while our conscious is busy with something else). The setting up of behavior patterns is the job of the conscious. It is done by trial and error. Formerly it appears to me, this trial and error was mostly directed by the emotions. Emotions being very rough abstractions of the success or failure of former actions in similar situations. These emotions are slowly replaced by "cold logic", a very approximate method for figuring out by rules from past experience what, this time, the future, i.e. the result of the action will be. So much for human thinking.
Now about "thinking machines". It seems to me impossible to know already how a mechanical brain will look, but at least the most fundamental parts, the principle, seem to me so intimately connected with any thinking process that they can already today be described. Using my definition of any thinking process: A thinking machine is a device to change the future to suit the owner. To do that it has to:
Part 1. Sense its surroundings.
Part 2. Receive the needs of the owner
Part 3. Determine the possibilities to fulfill those needs
Part 4. Act on these possibilities.
The job of part 1 will be: a) To sense its surroundings. b) To either identify these sensations with past similar ones or not. These sensations are coded into symbols, probably binary. Let us call them binary "numbers". lf identified, the old number is passed on, if not, a new number is given and passed on. c) All sensations, i.e. numbers are collected in a "reference frame," a sort of report, giving sensations and the places in space they came from. To include a time element always two consecutive reference frames are passed on as input.
Part 2 is by far the most interesting one. There the outside situation, the present, expressed as two "reference frames" - the collection of all sensations - and the needs, the wanted future, meet. They go to division a) The memory for all laws and rules. These laws and rules will consist of the scheme: situation - action - result; and may be specific or general. For instance it could read: door with handle closed - turning handle and pushing - door open. (All expressed in numbers). Or a general one: object in place - forceful movement applied - object moving in direction of applied movement. Here applicable rules are searched for under the numbers received. Probably they will not be found, because the input will be too specific or too general. So message is sent to b) the composition and decomposition of number concepts. There the number concepts are divided into their parts - a tree into a root, trunk, branches, foliage, et cetera - or parts into the whole, - stones, trees, rocks, snow, geometrical shape and size into a mountain, et cetera. All concepts are stored here as sequences A to B. This process goes on, until a general rule in the memory fits the wanted change of future. The one solution is sent to department c), where this one general rule is split up into specific rules, again with the help of a) and b). The specific rules are then given to Part 3. Here all necessary actions are a) collected, b) put into correct sequences, and c) done. These actions may be a moving of limbs or only typewriting.
- W. Fritz, I. Alfaro 314, Acassuso Prov. Buenos Aires, Argentina (old address).
(The editor, John W. Campbell Jr. )
An excellent and highly interesting definition of "thinking", both verbally in a phrase, and in terms of process, I feel.
For continuous reading, like a book - continue
Jump back to About the author /e-book Contents.
Copyright © New Horizons Press