Thursday, August 7, 2014

Abstract evolution - cybernetic, meta, cosmism. Turing machine and the Pure Apriori Conceptions. Emulation of universes and minds. Thinking machine vs Turing Machine - and much more... Continuation of the thread on G+. To be continued...

[Continues, see the previous posts]


Randall Lee Reetz

Yesterday 11:33 AM 
 
tho
What is evolution doing? What problem does it get better at solving? Come on people, it really isn't that complicated. Think!

...


Todor Arnaudov


Randall* (see the note in the end), one could just say: prediction, compression (shorter description for functionally equivallent items - it goes together with the former), preservation - however one  should add: "of what", for example of the "conserved cores" - see below. Also one should add increment of the range of prediction - higher precision, higher spatio-temporal range, higher certainty (see in the references as well). And the capacity for prediction, compression and preservation are encapsulated in spatio-temporal areas which grow both bigger and smaller - expanding resolution both in micro and in macro scale, and the encapsulated "areas" (sub-universes) get/aim at getting more and more independent from the rest of the Universe. They create higher forms of physical laws (causality), built of sequences/systems of the lower ones which predict correctly, and they get ever more aware of  them and certain in their execution (that's the improvement of prediction/compression, and increase of the quantity of "real causality" - that is one that causes in the external to the subuniverse lower Universe with the highest possible resolution of causality and perception, in terms of the target, lower level Universe). Etc.



All of these need elaboration and grounding, in order to be more than simple observations and claims, though, for example - see below.
I'm sharing some "science" on the topic of (abstract) evolution - by the way it's called "transhumanism", "cosmism", "cybernetic/abstract/meta-" evolution (by prior researchers) and yes - the ones who put humans (in their current form, emphasizing not systems', but the concrete biological substrate ) as the "ultmate top of the Universe" were religious ones, rulers of society (not the smartest one), and ones not seeing the more abstract sense, seeing only their narrow "self interst", if I could say, instead of truth. For instance Schopenhauer did say it 200 years earlier.

That subject was even taught/presented in the first AGI university courses ever (known to have existed), 3-4 years ago - of which I was the author.


There is the synopsys, the course program and the opinion of Ben Goertzel: http://artificial-mind.blogspot.com/2011/03/mathematical-theory-of-intelligence.html

There are slides (but mostly in Bulgarian: http://research.twenkid.com/agi_2011/
)

Vladimir Turchin on the Meta-System Transition - cybernetic view on the evolution: http://pespmc1.vub.ac.be/POSBOOK.html

Boris Kazachenko: http://meta-evolution.blogspot.com/

Todor Arnaudov: Theory of Mind and Universe ["teenage" is the time when it was conceived and first published by the author],
start with the introductory definitions about "Universe", "subuniverse".


Short presentation of some principles of the evolution of the Universe and intelligence in a lecture:
http://research.twenkid.com/agi_english/General_Intelligence_Principles_Caffe_Sci_2009_MTR.pdf

More theory and definitions:

Sides:
http://research.twenkid.com/agi_english/Todor_Arnaudov_Theory_of_Hierarchical_Universal_Simulators_of_universes_Eng_MTR_3.pdf

Some of the original works:

http://research.twenkid.com/agi_english/Teenage_Theory_of_Universe_and_Mind_4.pdf
http://research.twenkid.com/agi_english/Teenage_Theory_of_Universe_and_Mind_3.pdf

Discussion on "Entropica" and general intelligence:

http://artificial-mind.blogspot.com/2013/12/entropicas-equation-of-intelligence.html

Various; e.g. a discussion on ethical issues of transhumanism

http://artificial-mind.blogspot.com/2012/02/philosophical-and-interdisciplinary.html

I suggest also some recent notes on the Meta-Evolution regarding the conflict between preservation and progress and the materialized form of these processes in the bodies of male/female organisms of humans: (In Bulgarian, though)

http://razumir.twenkid.com/kakvomu_notes.html#233

That's it, male organisms are materialization of the "progress" arrow ("evolution"), female - of the preservation of the existing, the behavioral differences suggest it strongly, as well as the "settings" (males genome has evolved the basic female - the Y chromosome). Due to the various levels of "objectivation of the Will" (levels of evolution of the matter), it's not "flat".

On the "Definition of Machine Intelligence" by Marcus Hutter and Shane Legg (that's regarding the "bandwidth" problem as well) - slides taught in the AGI courses and the paper:
http://research.twenkid.com/agi_english/Machine_Intelligence_Hutter_Legg_Eng_MTR_Twenkid_Research.pdf

See also specifically the AIXI, their model for "Universal AI".

...

Well, the whole so called "dialectical materialism" is about abstract sense of the evolution, or as Engels/Lenin would probably say:

"Higher forms of motion of matter", where the abstract sense of "motion" is "dynamics", "change".

....

In general, you should define your terms distinctly and clearly, something that you don't, and you should understand and recognize common concepts in other participants or thinkers - Schopenhauers's "The Will as World and Idea" is also a theory about the abstract evolution and the production of higher forms of "objectivation" of the Will , where his concept "Will" includes aspects which are discussed below in the references.)


If you were not ignorant/lacking curiosity, you would have known, that Will is a special concept which refers to higher forms of Causality, the "On the fourfold root of the principle..." is also about that (1813). In my theory humans or all Causality/Control units, or Virtual universes at different levels - they are all higher forms of causality as well.


The "Pure apriori conceptions" of Kant are Time, Space and Causality - they map to Turing Machine's/Random Access Machines clock generator, memory and instruction set.

One important aspect is, though, that the Pure apriori conceptions are *empty* of content. They are not *empirical* (that's one other reason why there's nothing "scientific" in your claims; I saw you cited Goedel - logic and axiomatic mathematics is not empirical).


Instruction set should be defined precisely, there also must be specific material - configuration of the machine memory, otherwise it's dead.


An empty Turing machine does nothing - it is the SOFTWARE that's interesting and that has enough of material for further analysis.


Here's one of your other confusions - you "summarized" my first post so naively and wrongly - like if you were a dumb "keyword recognizer" - you noticed "emulation", therefore "Turing machine"...


However, in order to emulate even just one computer with another one, one Turing-complete device with another one (not a sensory-motor thinking machine, a versatile limitless self-improver, an AGI), you have to have a lot of other things which are much more complex (in any measure - having more structural complexity and depth) and more intersting that the Turing-completeness alone:


-- The software of the first
-- An EMULATOR
-- Have to transfer the software of the first to the second


 A Turing machine alone, without appropriate software and other means cannot write this emulator.

So I'm talking about the capacity to write the emulator, which requires capability to investigate and understand the other "Turing machines" - causality forces, sensory-motor data. That is, I'm talking not only about the "execution of instruction", but about understanding.

And yes, you could run AVX2 code on a 6502, but you first have to write the emulator, among the other requirements and the sufficiency of memory.

An empty PC or an empty Apple][ cannot do it, and if it does understand the instruction sets and knows how to write the emulator and make the mapping, it's obviously much more than just mindlessly "Turing complete".


Accordingly, my theory refers not to "Turing machines", but to Virtual Universes, to Cuasality-Control units, resolution of causality and resolution of perception; aims at defining causation in operationalized and quantifiable terms - for instance true/complete causation and virtual/conditional causation.


It talks about universal recognizers and simulators of virtual universes - that's related to what others try to model with Hierarchical HMM or with their AIXI/UAI, algorithmic probability; etc. etc.

(No, "subuniverse" in my terms is not "local minimum" - it has more structure. "Local maximum/minimum", in Calculus sense, is used however as an aim for the Causality-Control units; the different ones have conflictint aims, coordinates of maximum reward, that's why maximum for one is a minimum for another one and it shouldn't be defined without depth. You should refer to the provided works.


See also: http://artificial-mind.blogspot.com/search?q=analysis+of+the+meaning


Part 1 (и български):(This Post) Semantic analysis of a sentence. Reflections about the meaning of the meaning and the Artificial Intelligence

Part 2 (и български): Causes and reasons for human actions. Searching for causes. Whether higher or lower levels control. Control Units. Reinforcement learning.

Part 3 (и български): Motivation is dependent on local and specific stimuli, not general ones. Pleasure and displeasure as goal-state indicators. Reinforcement learning.

Part 4 : Intelligence: search for the biggest cumulative reward for a given period ahead, based on given model of the rewards. Reinforcement learning.

)


//// Furthermore the published works I cite are quite old and introductory, they need elaboration and more operationalization.)


Overall again: a thinking machine, an AGI, a Versatile limitless self-improver (VLSI) is much more than a mere Turing Machine - see more elaboration of the difference in the previous posts.


...


Now a huge difference that arises is the following:


An "empty", "reset" Turing machine does nothing. Void. It's in "HALT" state. Dead.

On the other hand, a mind or an "empty" VLSI is never in "HALT" state.
Human mind can't even imaging being in a "halt" state - it's just unconscious one, like if it didn't exist.

If a mind of a Versatile limitless self-improver doesn't have material, it starts to synthesize some and to scan and search for structure in the environment - just like babies, if put in dark, start to move their eyes and head around in a search for light/contrast difference - something to catch the attention and to allow further investigation, recognition, memorization, generalization, prediction, acting upon/with,  etc....



The Turing machine alone neither searches for anything, nor it knows anything.

The same applies for the Universe as a whole - it just does, it's the "blind Will", causal forces being applied.




And that's related to the thought I cited in the first comment:

"Where calculation begins, comprehension ceases"

Which refers also to the "Symbol grounding", to "Chinese room experiment", the discussion regarding behavior in "On intelligence" by Jeff Hawkins.

It refers also to the common lack of understanding of the concept of "grounding" (explained in the above posts of mine) from many people in the classical AI/programmers/logicians.

You don't seem to get it, even though I notice that you mentioned "building ontologies" and generalization, which is correct (and "ontology" is yet another philosophical term, in your "scientific" speech; your post about the "what of, what is" is poor/small scale philosophy as well).


...


A bit more of an expansion:

The capability to map correctly specific experiences/other input to known concepts/generalizations is called "Facculty of judgement".


It's related also to the so called "principle of homogenity and principle of specification" - generalization and specification.

It means, that you should generalize what's similar/common, but you should also accordingly discriminate and classify the differences in seperate and distinct classes.

If you do not do correctly, there will be confusions.


That's why you were criticized for the use of "evolution" with no specifications. That term alone was used and initially defined for *biological evolution*, which has its specifics.


Normally, the abstract or non-biological evolution has been called "Progress" - that's before Darwin, and it's not a new discovery or a concept - it was obvious in the advanced minds even in the early 19-th century (industrial revolution gave insights).


As of the more distinct elaborations on cybernetic evolution or the abstract evolution or meta-evolution or meta-system transition - see above.

That's "scientific" material - grounded claims, containing evidence, examples, definitions of concepts, references, trying to make predictions - so long as this topic is "science", because as "meta-" and "abstract" suggest - it's also philosophy - speculative and more abstract/general than science, inter-disciplinary, a super science. Just like "metaphysics" or "meta-programming".

...

Indeed your confused "parable" is sophistry, I missed the "scientific" part in your comments.
For example you share no *evidence*, just undefined and general dogmatic "claims", usually empty of content (like the Pure apriori conceptions).



BTW, If you try to refer to Ptolemy - he was actually a honest and convincing scientist (empirical), and he obviously has used empirical evidence.

At the time of conception Ptolemy's explanation was pretty persuasive - and no, it was before Christianity. The ancient world did have science and method, even the pre-school children, 4-5 year old are empirical scientists. Just their experience (sample data) and bandwidth is limited, so their hypothesis and theories are as credible and complicated as that allows.


See for example Jean Piaget dialogue with a child regarding what makes the wind blow. It's cited in the Developmental Psychology lecture in the course:

http://research.twenkid.com/agi_2011/Developmental_Psychology_30_4_2010.pdf

There it is for completeness:


Piaget: Who makes the wind blow?
Julia: The trees.
P: How do you know?

J: I saw them shaking their branches.

P: How that makes the wind blowing?

J (shakes her hends in front of her face): Like that. However they are bigger and there are a lot of trees.
P: Who makes the wind in the oceans?

J: It blows from the Earth. No. From the waves...

Even Piaget didn't get this right, he takes for an important aspect "Animism" - because the child sees the trees and nature as being impersonated, having intentions etc. - which in fact is not that wrong for two reasons:
1. There is no evidence/alternative theory at the time,
2. As many abstract thinkers agree, the Progress creates higher forms of causal laws thus human will/intentions and the lower forms of causality are kindred and members of a common class (see "On the fourfold root...")


Thus, to me there's something else that's more interesting: it's that the child does have *EMPIRICAL* facts collected by *EXPERIMENTS*, and then she makes inferences based on these empirical observations, which are correct and consistent.


-- Shaking hands produces wind (experimental settings, experiment, results)
-- The feelling of shaking hands in front of one's face is similar to the feeling when there's wind outside

-- The force of the wind can be quantified/measured and it depends on the magnitude of the surface being shook and the speed of shaking


On the next generalization:

---> Motion produces wind
* Branches of the trees with the leaves shake.
* Waves in the ocean move
* Bigger/wider objects cause stronger wind - as the wind in nature is stronger than at home

Therefore the trees and the waves produce the wind.

Respectively, if one looks at the *EMPIRICAL* data available to Ptolemy and the numbers/quantities known/in use by that time, one would see that he was a good scientists and didn't make wild guesses.


As of the process of discovering something beyond that - so that the other scientists (limited empirical observers with worse generalization capacities) can't be convinced with evidence...

The above requires a higher abstraction, to go beyond the competing thinkers, a wider scope.

 For example, the one who proposed that stars are probably far away, they are other suns with other planets and other living beings was classified as "philosopher" - was Giordano Bruno who died for it, killed by dogmatic idiotic apes.


And as it's always with the most advanced and the best minds - he was interdisciplinary, he was also a mathematician, astronomer (he extended Kopernik's model), a genius, unlike pedantic calculating machine.


Philosophy is the most general/abstract/speculative/meta- view on the rest of the knowlege and theory.


Philosophy appears always after science - and  your parables are nonsenses, don't mistake religious nonsense with philosophy, and don't use it as well... :)) That's why you can't provide ANY specific evidence, any name of a philosopher or a work, or a page in a treatise; or any historical fact that prove your nonsense - compare to the writings of the "philosopher".


To name something "philosophy", it must be a higher generalization on something else, it needs  material to generalize on, which is provided by empirical data, and as shown with example with the child - even 5 year-old has some form of a method for testing hypothesis, actually even a new-born has.


The only "sciences" without empirical data are such as Formal logic; some consider also pure mathematics - when it's based on purely axiomatic systems, that do not understand/relate/care/know their grounding. Mindless calculations which loses detail and ground - which for instance is the reason for the silly "logical paradoxes" (such as Goedel "mighty" theorem, proving an obivious - nature is not "formal logic". Some "scientists" still believe so, though.


However "nature never lies", Universe is always "correct" and it always "says the truth". If you find a paradox or can't predict/map/... it correctly/completely - it's your model/theory, not nature.

Important Note:


* As of the addressing to Randall - the purpose of this/previous posts is of course the (truth, depth, coverage, understanding, explanation, clarification, grounding, ...) of the subject matter. He's only an "instrument of the Will", "a soldier of fortune" as he himself notices about humans in the path of Progress (humans in the specific biological substrate) - a step in the whole process.


These writings with some more unpublished expansions and details, and in a better formatted presentation would go in a complete work.


I am aware that the specific "soldier of fortune" to whom formally the message was addressed (it's just formally, let me emphasize) does not have enough of "RAM,  CPU" or attention span to grasp  such elaborate studies as the above, so please let him do not believe that he's that special.


The topic is that deserves the attention, and some the peculiarities and confusions of minds like his are worth being explained.

No comments :