Expert: George Moustris - 7/12/2003

Question
AI, God. AI growth
---
I want you to discuss my idea of AI growth:
Suppose we have a set of information entities e1, e2, ..., e(n). Suppose if we want to generate new entities from them by combining several of them and creating resulting one: e(n+1) = e1+e2; e(n+2) = e1+e3;... If God exists, does it take time for him to process this information or does he know already everything (if it takes 0 seconds to create new entities from 2 existing). I mean, do you think God knows everything or still growing? This would be a good idea for AI to grow if it is "bored" - constant combining already known facts into new ones.
Examples: e1=John is a man; e2=All men die; e1+e2=e3=John will die. e4=e1+e2=John is a man and all men die. e5=e1+e3 = John is a man and he will die... Not all entities will form meaningful entities, but for AI it means constant growth, not so fast as God, but growth, faster then human intellectual growth. If we increase initial entities, AI can be very smart. It can understand such entities as "if John is black or white and Jack is black or Hispanic, then it is not true that John cannot be black and white or Jack cannot be black or Hispanic". Moore's law. It is not as evident for human, nut for AI it can be known fact.

Please answer it here if you like to discuss it farther:
http://pub152.ezboard.com/fscience10741frm9.showMessage?topicID=1.topic

Thanks.

Answer
I think what you mean as an information entity, is actually a logical proposition(and by the examples you give,they are just that).Combinations of logical propositions so as to produces new ones, is not a new idea.It is quite old(see Aristotle,Socrates or even older philosophers).Such combinations can and are made by machines.Even from the same old PC you use to write this posts.Back in the 50's the pioneers of AI thought they could program a machine to combine such propositions and virtually come up with all proposition a set of axioms has to offer(imagine geometry for example.The machine could combine all known theorems and prove new ones,and new ones and new ones...After some time it would have provided all the theorems geometry has).This was called Automatic Theorem Solver(or solver).Well,it didn't work!And had a good reason not to.It is called Godel's theorem.In plain terms this theorem says that given a well diefined system of propositions, there are some propositions whose validity(determing whether its true or false) cannot be proved.What this means is that you cannot prove whether some propositions are true or false!!They are undefined!(this is consistent with the theory of the nature of truth,but that is a very different story).
So, your theorem solver will face a fatal problem.It cannot determine whether the theorem it wants to prove,CAN BE PROVED or not(we wouldn't want our machine spending eternity trying to prove something which cannot be proved).
As for God now,what you say does not make sense since all empirical claims about God are false as there are NO empirical observations abour him.In other words no one knows what,where,who God is thus all statements about him are vague and speculations.Furthermore,if we accept the definitions of religion about God,then these definitions contradict with all our knowledge about nature.So *logically* they must be false.
Best G.