Wednesday, July 23, 2008


Artificial Intelligence (AI)
Andre Willers
23 July 2008

We trace the development of Artificial Intelligence .

We find that the requirements for AI is
A Basic Intelligence (see AI(0) below .)
An external Database
An Interface
An Unpacking system
A Packing system
See “ Unpacking” and “Orders of Randomness”

We see that humans in human cultures can be described as AI’s .
We see that numbers matter .
Cultures compete and evolve to incorporate as great a number of humans as possible .
This drives idividuals to a greater AI(number) .

AI(0) :
Our starting point is the unaided human brain . Unaided in the sense of an intelligent black box with no external aids like language , tools , others ,etc .
Define this as AI(0) .

AI(1) : Language
Language is both the External Database and the Interface .
Culture is unpacked and packed using a particular language .
The culture uses individuals as temporary memory .
Knowledge is stored in language and transmitted verbally . “The Word”
The whole acts as an AI .

Different cultures(languages) compete , especially during down-turns .
The winning languages unpack significantly different world-views .
The losers go to the wall in time-honoured evolutionary fashion .
They have fewer and fewer numbers of speakers , until they peter out .

A very good example is the idea of incremental change .
This is natural to a positional language like Latin or English , but not tonal languages like Chinese . It is easy to modify a position of a word , but not a tone without altering the sense altogether . A Newtonian worldview would not develop in a tonal-language culture .

AI(2) :Writing.
External Database : Stone , clay , papyrus , paper , etc
Interface : Eyes , Pen , chisel
Unpacking : reading
Packing : writing . It is interesting to note that non-alphabetic writing came first . The initial packing did not represent spoken language . The alphabet developed out of a pidgin-speak version of the Ancient Egyptian hieroglyphs . An interesting example of the primary evolution of unpacking systems .

AI(2.1) Writing – Eyes
The invention of mass-produced reading-glasses in 1298 AD in Northern-Italy enabled a mass-market of readers in the critical middle-aged , decision maker class . The number of AI(2) individuals increased exponentionally .

AI(2.2) Writing - Printing
This created a huge market , which the perfection of the moveable type printing press addressed .

The number of AI(2) individuals exploded , both numerically and as a percentage of the society’s population .

Cities exploded . The conventional view is that cities grew because of population growth . But even if population growth in the boondocks is negative , the cities still grow . The urge to a higher AI number for the grandchildren drives it . See previous posts on

AI(3) : Turing Computers.
(A Turing computer is one that does one thing at a time)

External Database : Binary data (punch cards , magnetic domains , etc)
Interface : Eyes (screen) , keyboard
Unpacking : reading and uncoding .
Packing : writing and coding .

This is familiar to all of us (2008) .
The external database has programmable features accessed by coding . This is quite powerful .

AI(3.1) Direct Neural Interface
Analogous to eye-glasses in AI(2.1) , a direct neural interface to a Turing external database creates a larger class of informed decision-makers .
This is especially useful in cases where we have a theoretical background , but the sheer mass of data to be processed in near-real time hampers AI(2) systems .

This is a definition of economics .
This is where we see the first application of it as well .
Direct Neural Interfaces came from the “Look-and-shoot” helmets developed in the US towards the end of the Vietnam war . Various signals using biofeedback training were used . The research went classified and surfaced as an advisory service in the Reagan administration (1982) .

It’s been running the show since .

A note on secrecy:
Up till the WWII , technology developed in the private sector was adapted and further developed by the Military-Industrial complex.

Two things changed it : Manhatten and Ultra .
It is about allocation of resources . Allowing oversight by short-term committees will destroy promising lines of enquiry . The only way to keep the money flowing is by secrecy . Of course there is waste . That is inherent in the nature of any inquiry where the end is not known . Experimental science means waste . But the successful payoff compensates for all losses and more . Failure can mean extinction .

Nowadays , basic research is done by the Military-Industrial complex , and spin-offs get privatized . A good example is Oracle , a CIA dbase project that got privatized by permission to Larry Ellison .

Or the Internet , a DARPA creation .

The irony is that the Military-Industrial complex itself is changed by this process . Functions previously performed by the academic world gets dumped on them . Since they are by definition more threat-conscious (ie paranoid) than academia , they are forced to explore low-probability pathways . But there is a finite tax-base .

The slow destruction of tenure in Western Universities is a ripple effect of the shift of resources to the Military-Industrial complex .

The Military-Industrial complex and any AI has a vested interest in an increased tax-base . More people and richer people .

Has this happened before ? Of course , many times . But attention is drawn to the development of large war-horses during the later part of the Roman Empire , and the feudal knight-church castle combination that evolved out of the western roman empire.

But hopefully we can do better than that .

To paraphrase Churchill :
“War is too important to be left to the generals.”
“Democracy is too important to be left to the people. “

What is it like to be a AI(3.1) ?
See “What is it like to be a fish?”
The first pioneers must have had a torrid time . I salute them .
But routes get mapped . Newcomers get shown the ropes by veterans .
And so it goes .

At first it like sticking your head into bubble where at least ten heavy metal rock-bands are playing simultaneously . With a guide , the volume can be decreased and the data streamed .

It does make you smarter , but not excessively so .Where there is already a theoretical background that can be filled in by large volumes of data (like economics) , it works very well .

But alas , nobody becomes godlike , not even weakly .

What the brain does learn is clumps of mirror-neurons. The survivors of this process learn to pack more information into the mirror-neuron complex than it can contain , except for a special unpacking complex . There are quantum-processes involved .The existence of the unpacking complex means that the packing system can exist .

AI(4) Quantum computers

4.1 Person Upload
A brain-body can be uploaded into a quantum computer using existing technology .
Simply use quantum teleportation into a quantum computer .
We still have the AI(0) brain , plus interface and external database.
Theoretically , we can dial up any level of AI up to level 4 . I have a suspicion that this would be easier said than done .

This can be done now (probably has been done) .
But this is done on a human template .

I draw your attention to what happens after puberty in humans when the neural mass increases significantly .

As expected , computational capability increases . To be more exact , more mirror-neuron-complexes are formed . Each one runs through it’s time- expansion .
They then vote on a course of action . The result of this vote is not a foregone conclusion , as any parent knows .

This means that any AI(4) is totally unpredictable from AI(3) levels .

AI(4) can only be constrained by AI(4) and above . The reason for this should be clear to anybody who has paid any attention at all to my previous posts on dimensional generation .

4.2 Real AI(4)
Can a real AI(4) be built by a AI(3) ?
Yes , by using Beth(2+) systems .
Will this be divorced from human antecedents ?
Yes , it is possible .
It is not possible using beth(0)
What does this mean ?
Self aware robot slaves are unreliable .
Not a good idea .
If you have to create AI(4)’s , use human templates . Your survival rate will be much better .


No comments: