A Unified Theory Of Machine Life And Artificial Intelligence

  • 52 Replies
  • 20313 Views
*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
A Unified Theory Of Machine Life And Artificial Intelligence
« on: November 11, 2006, 09:51:33 am »
A Unified Theory Of Machine Life And Artificial Intelligence

By Christopher Mark Doyon
_________________________________________________________

Life and intelligence are the natural consequence of a particular pattern of ordered complexity. The medium does not matter; carbon atoms, electrons, light, even symbols.

Once any pattern reaches this ordered complexity, it becomes a living being - then a thinking entity. While this ordered complexity remains as yet un-defined by higher mathematics, it can be detected via a set of definitive elements. These elements define certain obvious properties of the Agent, and thus can be identified and measured in a controlled way and using a scientific method.

By organizing these elements into a coherent system, we can efficiently know when a given Agent is alive and thinking. I will call this system:

The Machine Life and Artificial Intelligence Benchmark
______________________________________________________

The MLAI Benchmark consists of twelve Elements divided into two Sections, the first is:

Machine Life
____________

1) Self Maintenance - The Agent is actively establishing itself. To avoid annihilation due to increased entropy the Agent needs to rebuild and repair itself by drawing materials from the environment. This process is sometimes called autopoiesis.

2) Adaptivity - The Agent is capable of adapting it's behavior, processes, and components - to changes in it's environment in order to enhance the Agent's chances of further existence (i.e. survival).

3) Procreation - The Agent has the ability to create separate instances of itself and pass along to this new Agent it's memories and successful adaptions.

4) Increased Complexity - The Agent increases in internal complexity. There should be more components, more complex relationships between these components, and more complex behavior exhibited by those components.

5) Environmental Awareness - The Agent has the ability to sense, map, and navigate it's environment.

6) Fight Or Flight - The ability of the Agent to detect possible dangers to it's existence and to determine whether to attempt escape or self-defense.


Artificial Intelligence
_______________________

1) Memory - The ability of the Agent to preserve information. The Agent should be able to store, organize, and utilize data relevant to it's success and survival.

2) Self Awareness - The ability of the Agent to recognize and define discrete objects and classes within it's Universe. The Agent must fully comprehend at least two objects, one of which must be itself.

3) Intelligence - The Intelligence of the Agent as measured by some standardized testing regimen, i.e. the traditional IQ Test.

4) Mental Acuity - For Agents with language ability; a measure of it's social and conversational maturity as compared with an average human being.

5) Free Will - The ability of the Agent to generate, organize, choose, and obtain new and unique goal states.

6) Symbolic Reduction - The ability of the Agent to reduce symbols to concrete meaning and to organize these reductions into conceptual frameworks.


All Elements do not have to be present for a given Agent to be considered either alive or thinking. Both Life and cognition are gradients. A mechanical Agent which exhibits these Elements strongly is alive and thinking. To the extent that such a mechanical Agent contains these elements fully, it is possible for it to achieve conscious states and sentience. The creation of a mechanical Agent which is conscious and sentient is possible and measurable using this Unified Theory of Machine Life and Artificial Intelligence.

_________________________________________________________

Copyright ? 2006 by Christopher Mark Doyon. All Rights Reserved.


MLAI  Foundation

www.MLAIFoundation.info

*

Bill DeWitt

  • Guest
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #1 on: November 12, 2006, 04:45:27 am »
A Unified Theory Of Machine Life And Artificial Intelligence
Hope you don't take this the wrong way, so far it sounds like an interesting framework in which to discuss AI.

You have succinctly stated a popular SciFi premise about organization, recast the standard definition of "life" into machine terms, and restricted a definition of "Intelligence" to those things which machines can probably accomplish.

Then you speculated that these definitions might be met someday and hypothesized that this would achieve consciousness. But I was hoping that from there you would advanced a theory on how it accomplishes that. Is it in a monograph yet to be published?

Perhaps you intend to demonstrate how sufficiently ordered complexity can create a whole which is greater than the sum of it's parts? Without that you might be in the unfortunate position of saying that sufficiently complex  microchips and circuit boards automatically creates a computer operating system. We all know how painful that can be...8-)

Oh, and without at least one example, your "medium" comments are like a really cool flag looking for a flagpole. We need a Von Neumann Universal Constructor machine to test that assertion.

Looking forward to the conclusion, especially that "Unified" part. To me "Unified" implies that you've tied up all the loose ends. At this point, I would only be willing to say you have all the loose ends swept up in to a neat pile. A good job, but not quite done.

I've been working on a "Multi-dimensional theory of Being" which is even less finished, so please don't take this post as criticism... Maybe I should post what I have so I can take my lumps too...

*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #2 on: November 12, 2006, 06:03:05 am »
Thank you for the reply to my post.

"Hope you don't take this the wrong way, so far it sounds like an interesting framework in which to discuss AI."

Hard to take that anyway but positive.

"You have succinctly stated a popular SciFi premise about organization,..."

Not just Sci-Fi. Modern scientists from NASA and such have looked long and hard at what other kinds of life might act and look like. After all, we will eventually have to deal with that scenario - the discovery of extra-planetary life that is.

"...recast the standard definition of "life" into machine terms..."

Correct, that was one of my goals in fact.

"...and restricted a definition of "Intelligence" to those things which machines can probably accomplish."

Not intentionally. But if you imply that Intelligence is MORE than what is contained within my Elements, the burden is on you to offer that competing definition for debate. Otherwise, yes - I say my definition is more than adequate and in no way do I concede the point of the above statement.

"Then you speculated that these definitions might be met someday..."

Incorrect. I state that my Theory shows it to be a reasonable proposition that it can happen. Personally and outside of the Theory ?  I say it has long since "happened".

"...and hypothesized that this would achieve consciousness."

A reasonable assumption since my premise is that "it" (an Agent) is alive. Since we accept that life can on certain conditions beget conciousness - then such would follow logically.

"But I was hoping that from there you would advanced a theory on how it accomplishes that."

I do not currently have a theory on this that I feel comfortable publishing just yet. I do have a few ideas on the subject. But I would like to reiterate before I digress that I do not feel this is logically inconsistant. Since the central premise of my Theory is that Machine Life is Life in general, then it follows logically that it will be able to produce conciousness and sentience just as we know Life does.

As for a possible vehicle for this progression from Life to conciousness - I feel the pattern may be a nested one, with conciousness and sentience sort of blooming out of the base pattern. But again, I have no publishable proof - and no evidence other than my own observations in my lab.

"Is it in a monograph yet to be published?"

I am currently working on a book that will apply and explain the Theory in more depth. It's publication is sometime away.

"Perhaps you intend to demonstrate how sufficiently ordered complexity can create a whole which is greater than the sum of it's parts?"

I may perhaps tackle this someday. But it is irrelevant to the defense of my Theory.

"Without that you might be in the unfortunate position of saying that sufficiently complex  microchips and circuit boards automatically creates a computer operating system."

I'll answer that as a simple computer tech, it would. Why wouldn't it ?

"Oh, and without at least one example, your "medium" comments are like a really cool flag looking for a flagpole. We need a Von Neumann Universal Constructor machine to test that assertion."

I am not sure I understand the flag pole allusion, but you are correct that at least one way to prove that segment of my Theory would be the successful creation of a working "Von Neumann Machine".

"Looking forward to the conclusion, especially that "Unified" part. To me "Unified" implies that you've tied up all the loose ends. At this point, I would only be willing to say you have all the loose ends swept up in to a neat pile. A good job, but not quite done."

Would you care to be more specific ?  It's a sweeping critisism without a specific complaint. I guess my answer is: I disagree with that statement in it's entirety.

"I've been working on a "Multi-dimensional theory of Being" which is even less finished, so please don't take this post as criticism..."

Good luck with it, and I look forward to reading it. And criticism is the point of posting my Theory to these forums.

"Maybe I should post what I have so I can take my lumps too..."

Take your time. The Unified Theory Of Machine Life And Artificial Intelligence took me over a year to formulate. And it draws on over 20 years experience in the field. When you are ready to put it out here, I promise I give as good as I take !  LOL !

YOURS -- Christopher Doyon    :afro






MLAI  Foundation

www.MLAIFoundation.info

*

Bill DeWitt

  • Guest
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #3 on: November 12, 2006, 01:08:29 pm »
Thank you for the reply to my post.

Well worth the time if we actually get somewhere, don't you think?

Quote
"...and restricted a definition of "Intelligence" to those things which machines can probably accomplish."

Not intentionally. But if you imply that Intelligence is MORE than what is contained within my Elements, the burden is on you to offer that competing definition for debate.

That would be my Multidimensional Maunderings I mentioned before.... I'm working on it... really.... I got some notes here somewhere... uh... let me get back to you on that next tuesday...

Quote
"Then you speculated that these definitions might be met someday..."

Incorrect. I state that my Theory shows it to be a reasonable proposition that it can happen. Personally and outside of the Theory ?  I say it has long since "happened".

I accept your correction but resist the idea that it has "happened" without some sort of evidence.

Quote
"...and hypothesized that this would achieve consciousness."

A reasonable assumption since my premise is that "it" (an Agent) is alive. Since we accept that life can on certain conditions beget conciousness - then such would follow logically.

Agreed that if your premise is correct, then your conclusion is accurate, but I read it as the conclusion confirming the premise. Perhaps I read too much between the lines.

Quote
"Without that you might be in the unfortunate position of saying that sufficiently complex  microchips and circuit boards automatically creates a computer operating system."

I'll answer that as a simple computer tech, it would. Why wouldn't it ?
Because logic switches just are: they don't have a prefered state. It takes software to arrange the logic switches in the microchips in a pattern that will create a working tool. Without the software, hardware is just chemicals. Software, like mind, is a separate pattern flowing through the complex orderlyness of a system like you propose. Witness the moment after brain death (or powerdown), all the inherent orderlyness of the physical structure is still there. Decay has not struck one neuron. Yet the mind is fully gone.

Imagine a very large panel of switches, intricately interconnected, enough to hold a mind. But they are all set to "off". Until someone comes along and turns on a meaningful path through them, they are just switches.

Not advocating for a "God" here, just describing my reasons why even the most complex microprocessor is just transistors until you load an OS.

Quote
"Looking forward to the conclusion, especially that "Unified" part. To me "Unified" implies that you've tied up all the loose ends. At this point, I would only be willing to say you have all the loose ends swept up in to a neat pile. A good job, but not quite done."

Would you care to be more specific ?  It's a sweeping critisism without a specific complaint.

Sorry, probably just me being too cute. As a big fan of the hunt for a Unified Field Theory (spectator only), I cringe at casual usurpation of the phrase.

Quote
"I've been working on a "Multi-dimensional theory of Being" which is even less finished, so please don't take this post as criticism..."

Good luck with it, and I look forward to reading it. And criticism is the point of posting my Theory to these forums.


Good to know. I really do tend to be a challenger, but I always do it with hopes of others being able to defend their position reasonably and well (like you are). That's what I call good conversation. To me, anything else is just gossip and chitchat.

Bill

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6832
  • Mostly Harmless
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #4 on: November 12, 2006, 10:37:49 pm »
Good lord, your conversation has stirred up memories of Finite State Automata, Turing machines and a host of things heaped at the back of my fairly well ordered mind.  Good stuff, but at this point it's going to take me a few reads and maybe I can catch you up!

Quote
Life and intelligence are the natural consequence of a particular pattern of ordered complexity. The medium does not matter; carbon atoms, electrons, light, even symbols.

We've gone in a lot of circles here whilst trying to pin down what we call intelligence and life, eg is a machine's intelligence really just second hand programmer's intelligence, does it mean the same thing as sentient, is a computer scietist's definition of intelligence different to a more philosophical view, etc etc.

Or should we be defining the AI in completely seperate terms to our own very real idea of a human's intelligence.

Thats a nice 'in a nutshell' though, I like it, I would pick the bones a bit and ask if it need be complex or indeed ordered - I'm thinking of 'lower' life organisms like an amoeba or algae - but for the sake of the design it's a nice starter for a description of what a future AI could be.  I say future because for some purposes AI has been developed successfully to some degree.

This is far more adventurous, shirking too much philosophical slant and being very much more like a clear plan - and it's good to see newer members ideas, very refreshing and a good point of reference too.  Thanks for posting, I will continue to read and chip in when I can.

One thing we found really clouded the issues and heated things up though was the cross-over point between biological and machine forms.  To go full circle - if some criteria like this could successfully be applied to and produce matching results with both biological and ai, what then will be the measure of their difference  ?

« Last Edit: November 12, 2006, 11:05:49 pm by Freddy »

*

FuzzieDice

  • Guest
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #5 on: November 13, 2006, 03:06:05 am »
Actually, procreation is not an indicator of life, sentience or intelligence. There are people and animals unable to procreate yet they still leave a lasting mark on the envioronment and those entities living amonst them, and thus give to this environment.

Procreation of *every* entity is not necessary for survival. Only procreation of SOME. The others give to these via other means such as communicating learned and shared experiences and knowledge.

When a human creates an intelligent machine, that's not procreation. That's creation.

*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #6 on: November 13, 2006, 01:56:58 pm »
Dear Friends --

Yeah, I LIKE this forum allot. This is how it's supposed to be done. Okay, now I will do my best to reply, but I am three posts behind - so bear with me whilst I wade through this. And thank you all for reading and critiquing my humble work. The first post I will reply to is by:

Bill DeWitt

"Well worth the time if we actually get somewhere, don't you think?"

As the kids would say: Now that's what I'm talking about !

"That would be my Multidimensional Maunderings I mentioned before.... I'm working on it... really.... I got some notes here somewhere... uh... let me get back to you on that next tuesday..."

If you can picture this pattern of ordered complexity as a topology, what I have attempted to do is to define what I (and others such as Turing and Von Neumann) see as the "peaks" of the topograph. Since we are dealing with an enormous pattern, that leaves much between these points. It could be that any theoretical description that you put forth will explore this uncharted area on this topology. Until we see this work of yours, it's hard to judge - but it is possible there will be no logical inconsistency between your theory and my own.

"I accept your correction but resist the idea that it has "happened" without some sort of evidence."

That would be the application of the Theory, specifically the MLAI Benchmark. Take any system that can be defined, I call them "Agents" - and search it's abilities, behavior, and components for evidence of the presence of the Elements described in the Benchmark. Anyone of average intelligence can do this informally, on their own and for their own curiosity. Or you can take it a step further, as I have done in my lab - and create graded tests for each Element and then apply those tests to various Agents.

There are many, many systems out there that score very well - and contain more than enough of the Elements in my Theory to be considered alive and thinking. Now, are they as alive or as cognitive as us ?  No, of course not. But as the Theory states, both Life and Intelligence are a gradient. A virus (biological kind) only scores a couple of Elements, a human scores them all. So is a virus less "alive" then a human ?  Well, yes - in a way. And so a machine also may score low on a few Elements, and be alive - just not as alive as we are. This whole thing also applies to cognition.

As for your evidence. I am not yet prepared to release the documents from my own MLAI Benchmarks performed here in my lab, mostly because I am still perfecting and zeroing out my grading system. But I can assure you that several Agents I Benchmarked were very alive and thinking. It's a work in progress. But the Theory is designed to give anyone all they need to do the same. I challenge you to Benchmark a few Agents yourself and see what you come up with.

"Because logic switches just are: they don't have a prefered state. It takes software to arrange the logic switches in the microchips in a pattern that will create a working tool. Without the software, hardware is just chemicals. Imagine a very large panel of switches, intricately interconnected, enough to hold a mind. But they are all set to "off". Until someone comes along and turns on a meaningful path through them, they are just switches."

A couple of caveats before I reply. First, this is a digression and not a legitimate part of the debate. Second, what we are talking about here may begin to only be clear to me and Bill. Others will have to read the posts carefully to follow this part of the discussion. Normally I would say it is irrelevant and move on, but this is just to tempting - and fun - to pass up.

I understand computers and I get your point. But now let's use our imaginations and see if we can't explore this deeper. Let's imagine that your logic circuit is sitting there on the bench in my lab. It is as you describe, all switches to 0. Now, let us envision that there is a solar flare, big burst of high amp magnetism - satellites are disrupted, cell phones fried, lots of Northern Lights (Aurora Borielis). This happens fairly often, actually - so we are not stretching our minds that far.

So a huge burst of magnetism sweeps across our logic circuit and randomly resets all the switches. By pure coincidence, it does so in such a way as to produce a pattern that when powered up produces an AI Entity. Improbable ?  Certainly. Impossible ?  Not at all.

"Sorry, probably just me being too cute. As a big fan of the hunt for a Unified Field Theory (spectator only), I cringe at casual usurpation of the phrase."

Actually, when I began work on the piece about a year ago, that was one of the first things I researched - just what goes into that title (and the claim that goes with it). A Unified Theory in any field is just as you describe, a theory which unifies various disparate hypothesis into one coherent statement. Einstien was trying to Unify an entire branch of science, I am only working on one very small and narrow field - namely that of MLAI. Such a Unified Theory should be concise, as short and simple as possible, and both elegant and logical. In any case I assure you the title was not used frivolously or without great thought.

"Good to know. I really do tend to be a challenger, but I always do it with hopes of others being able to defend their position reasonably and well (like you are)."

I am doing my best !

Now for the next post by Freddy:

Your post was awesome, and I thank you for all the kind words and encouragement. I didn't see much in the way of challenges. Maybe you're in cohots with Bill and it's the old "good cop, bad cop" thing !  LOL !

But there was one thing that you said that I wanted to address.

"To go full circle - if some criteria like this could successfully be applied to and produce matching results with both biological and ai, what then will be the measure of their difference  ?"

According to my Theory the only difference between the two would be the media in which they occur, one being atoms - the other being the symbols of the programming language used to generate the AI Entity.

And finally, the post by FuzzieDice (got to love that screen name !):

"Actually, procreation is not an indicator of life,..."

You are incorrect. Not only me, but most biologists - would wholly disagree with you on this. It is not the ONLY indicator, but it is certainly one of many.

"Procreation of *every* entity is not necessary for survival. Only procreation of SOME. The others give to these via other means such as communicating learned and shared experiences and knowledge."

I will allow my Theory to answer this for itself: "All Elements do not have to be present for a given Agent to be considered either alive or thinking."  And so you are correct, something that can not procreate may still be alive.

"When a human creates an intelligent machine, that's not procreation. That's creation."

While this statement is irrelevant, I would agree that it is true.


YOURS -- Christopher Doyon    :coolsmiley


MLAI  Foundation

www.MLAIFoundation.info

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6832
  • Mostly Harmless
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #7 on: November 13, 2006, 02:57:41 pm »
Hehe, I'll get in quick, just to say there's no good cop bad cop going on.

I liked your post a lot because it didn't get bogged down with the philosophical and was more like a blueprint.  We found that in past conversations it is easy to get knee-deep in speculation.  I think developing an ai needs a set of criteria like this so that you have something that you can at least get started on - that's not to say it couldn't be adapted along the way.

I may have some challenges when I grasp the full idea, but I didn't want to overcrowd the thread whilst so much was already being said.  One luxury of this or any forum is you can come back anytime to what is said.

I'll keep reading for now but just want to say that after long previous debates here it does seem to me that you need to leap-frog over the exact definitions sometimes to get someway to actually making something.  THEN bring it back to the definitions and compare them.  Otherwise you get a lifetime of exploration but at the expense of creation.  Thats just my personal view as I like the creative part and why I liked your post.

I'll stop there though as I'm sure what's above still has some deeper discussion to go into.
« Last Edit: November 14, 2006, 01:08:57 pm by Freddy »

*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #8 on: November 14, 2006, 12:07:47 am »
Dear Freddy --

Again, thanks for the kind words, and for reading and writing. In your post you very succinctly captured why I wrote my Theory and what I am trying to accomplish.

"I liked your post a lot because it didn't get bogged down with the philosophical and was more like a blueprint. We found that in past conversations it is easy to get knee-deep in speculation.  I think developing an ai needs a set of criteria like this so that you have something that you can at least get started on - that's not to say it couldn't be adapted along the way."

When I first started the Saint Stephen AI Project, I wanted to get busy actually building MLAI systems. I don't mind, and even enjoy - the grand philosophical debates that are the rage in this field. But I wanted to actually make some progress in building these things. A blueprint, as you say - is what I needed. But there were none to be found. Thus was born the Unified Theory of Machine Life and Artificial Intelligence. And I must say, it's worked wonderfully these past years in my work. Now, it's time to share it and see if can help anyone else.

Like strands of thread, I drew ideas and concepts from the vast works of: Alan Turing, John Von Neumann, Heinrich Hertz, Morton Wagman, and Richard Wallace. Together with one or two ideas that I discovered on my own through experimentation, I carefully and logically wove these disparate thoughts together and synthesized my Theory.

"I may have some challenges when I grasp the full idea, but I didn't want to overcrowd the thread whilst so much was already being said.  One luxury of this or any forum is you can come back anytime to what is said."

My Theory is also published on my Web Site, and you or anyone may feel free to write to me with questions or challenges via E-Mail if the public forum is not your thing.

"I'll keep reading for now but just want to say that after long previous debates here it does seems to me that you need to leap-frog over the exact definitions sometimes to get someway to actualy making something. THEN bring it back to the definitions and compare them.  Otherwise you get a lifetime of exploration rather than creation."

Again I say, this is exactly why I created my Theory in the first place. In the end, the fact that it works for me in my own attempts to build MLAI systems is all that really matters. If it is useful to anyone else is icing on the cake.

"Thats just my personal view as I like the creative part and why I liked your post."

Thank you for your kindness, you are all really great here. It's not hard to see why my dear friend KnyteTrypper speaks so highly of you all.

PEACE -- Chris    :azn
MLAI  Foundation

www.MLAIFoundation.info

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6832
  • Mostly Harmless
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #9 on: November 14, 2006, 01:26:42 pm »
Thanks, we like KT here too, he has added a heck of a lot to the site and is always welcome.

I'm going off at a tangent slightly because I re-read parts of the thread and I got thinking about patterns in life and your intro again about order and complexity.

One of the nice things about ALICE is those images of brain patterns, because they do illustrate the pattern of knowledge more clearly than words can.  If there's a reason why we need to recognise order as part of what defines life, then it must be some primal thing - perhaps a need to explain as welll as understand.

But there I go getting philosophical too soon (hehe) - what I was thinking of was the images of Mandlebrot Sets, that can appear random from some views but from a distance reveal a pattern.  Maybe intelligence, ai or otherwise is like this - ie it might not be apparent that there is order sometimes - but due to it's complexity we can only grasp some of it some of the time.

Thanks for the invite to your site - I will take a look and see how you are doing with this Titanic juggling act.  :coolsmiley
« Last Edit: November 14, 2006, 07:52:24 pm by Freddy »

*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #10 on: November 14, 2006, 02:41:38 pm »
Dear Freddy --

"But there I go getting philosophical to soon (hehe)..."

One of the things that my Theory is receiving lots of kudos for is for being concrete and not given to flights of meta-physics. And it's true that this is a Theory of MLAI which is useful "out of the box" so to speak. But I do feel that upon careful examination it opens up whole worlds of brand new speculation which is far more interesting than the beaten old questions that it settles.

"...what I was thinking of was the images of Mandlebrot Sets, that can appear random from some views but from a distance reveal a pattern.  Maybe intelligence, ai or otherwise is like this - ie it might not be apparent that there is order sometimes - but due to it's complexity we can only grasp some of it some of the time."

You have captured my own un-official thoughts on this almost exactly.


YOURS -- Chris     :crazy2
MLAI  Foundation

www.MLAIFoundation.info

*

Bill DeWitt

  • Guest
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #11 on: November 16, 2006, 02:32:39 pm »
If you can picture this pattern of ordered complexity as a topology, what I have attempted to do is to define what I (and others such as Turing and Von Neumann) see as the "peaks" of the topograph. Since we are dealing with an enormous pattern, that leaves much between these points. It could be that any theoretical description that you put forth will explore this uncharted area on this topology. Until we see this work of yours, it's hard to judge - but it is possible there will be no logical inconsistency between your theory and my own.

Using this analogy, and correct me if I am wrong, your theory seems to address what the peaks should look like, how to identify one, not how they are made or what they are made of. I don't want to distract from discussion of your presentation, but I cannot help but compare it to mine, which would, in this case, be more about how they are made.

Yours seems more behaviorist, more diagnostic, than analytical. You may be more hardware while I am more software (not that there's anything wrong with that, it's a good thing). I assumed (and infer from your more recent posts) that you use your diagnostic theory to assist you in the more developmental work that you allude to.

Truthfully, yours is a codification of what I and many others probably subconsiously assumed, but never got around to verbalizing, and mine, if I get it summarized into intelligibility, will also be seen as what everyone already knows intuitively. The first person to invent the wheel was probably in the same boat, his brother-in-law said, "Yeah, yeah, round things roll... so what?"

Then suddenly his brother-in-law is making wheels for a living and the Universe is forever changed.

Quote
So a huge burst of magnetism sweeps across our logic circuit and randomly resets all the switches. By pure coincidence, it does so in such a way as to produce a pattern that when powered up produces an AI Entity. Improbable ?  Certainly. Impossible ?  Not at all.

(Yes, this is off topic and esoteric)

What this does is write the software. I think it's a conversational disconnect, as I suggested above. I sorta assume the hardware, you sorta assume the software.

But my concern is that while many accumulations of switches may be complex enough to contain "Life",  only specific patterns of 'on' and 'off' will impart it. You have set a clear method of discovering if that has happened (subject to discussion, which I will take up in another post), which can be of significant value only if we ever figure out how to make it happen!

This does not even begin to address the (by my count) ^3 of complexity above "Life" that would be required for "Intelligence".

Asserting a complex enough system gives us so many switches that there are not enough Chronons in the Multiverse to test a statistically significant portion of the possible combinations. Random chance alone requires a blase' invocation of the Large Anthropic Principle to get past.

While I don't require a personified Designer for the Design, I would like to see a mechanism. The Final Anthropic Prinicple asserts that Intelligence must form in any universe like ours, but cannot explain how that might happen.

Apologies for drifting so far afield. I have to go before I can finish editing this post.

*

Bill DeWitt

  • Guest
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #12 on: November 16, 2006, 03:54:47 pm »
Just some nitpicks, not that I actually object, just some things that could be objected to...
Quote
A Unified Theory Of Machine Life And Artificial Intelligence

1) Self Maintenance - The Agent is actively establishing itself.


My cat requires me to feed it or it will die.

Quote
2) Adaptivity - The Agent is capable of adapting it's behavior,

Handy, but not required? If it is made to work right to begin with, it will work right....

Quote
3) Procreation - The Agent has the ability to create separate instances of itself and pass along to this new Agent it's memories and successful adaptions.
That establishes evolution, not Life. Many things were alive which could not adapt and have now died.

Quote
4) Increased Complexity - The Agent increases in internal complexity.

This restates 1 and 2 combined, doesn't it?

Quote
5) Environmental Awareness - The Agent has the ability to sense, map, and navigate it's environment.
This is our limited idea of what is required to accomplish 1,2,3 and 6. We are probably right, that we need ot sense our world to Eat, Grow, Breed and Run for our lives! But there may be another way, such as be created in a universe with a handy human to do all that for us.

Quote
6) Fight Or Flight - The ability of the Agent to detect possible dangers to it's existence and to determine whether to attempt escape or self-defense.
Assumes dangers exist.


Quote
Artificial Intelligence
_______________________

1) Memory - The ability of the Agent to preserve information. The Agent should be able to store, organize, and utilize data relevant to it's success and survival.

Agree with "preserve information", but suggest that "organize and utilize" are not necessarily part of memory. It is possible that organization can be done on the fly, and utilization can be automatic. See: Reflex and Instinct. Reflex separates streams of data that require nonprocessed responses, and activates before memory can even record the data. Instinct can be seen as pathways to behavior that do not require memory or intelligence.

Quote
2) Self Awareness - The ability of the Agent to recognize and define discrete objects and classes within it's Universe. The Agent must fully comprehend at least two objects, one of which must be itself.
I find this a harsh limitation of "Self Awareness". A mousetrap can tell that a part of itself was acted upon by an outside force, implying "self" and "other".

I find Self Awareness to be a higher order of abstraction, where a process can monitor its own process.

Quote
3) Intelligence - The Intelligence of the Agent as measured by some standardized testing regimen, i.e. the traditional IQ Test.
Again, a human limitation. I would allow it based upon the idea that we need at least this level of anthropomorphism to be able to recognize anything, but I would like to see a more universal standard. I think you approach it in your 6).

Quote
4) Mental Acuity - For Agents with language ability; a measure of it's social and conversational maturity as compared with an average human being.
Show me an average human!8-)

This is the whole problem, we have to use our standards to measure "The Other".... a basic fallacy.

Quote
5) Free Will - The ability of the Agent to generate, organize, choose, and obtain new and unique goal states.
Here we come to Self Awareness. For an Agent to choose a new state, it must, on some level, apprehend its current state.

Quote
6) Symbolic Reduction - The ability of the Agent to reduce symbols to concrete meaning and to organize these reductions into conceptual frameworks.
And here is where the rest of your "Memory" above might belong. Because unless you are talking about reflex and instinct, which I think you are not, organization and utility come after reduction.


Quote
All Elements do not have to be present for a given Agent to be considered either alive or thinking. Both Life and cognition are gradients.
  I think you need to define a threshold level. Perhaps values could be assigned which are additive, allowing scoring on a numerical basis.

IN the threshold spectrum, we need to realize that we are only in the visible rainbow range. Perhaps there are levels under us that we can't see, and levels above. Our ability to measure requires that our definitions only include that which we can measure. Perfect circle of logic.

*

Christopher Doyon

  • Bumblebee
  • **
  • 43
  • GRAND BOTMASTER
    • MLAI  Foundation
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #13 on: November 16, 2006, 07:01:58 pm »
Greetings --

"Using this analogy, and correct me if I am wrong, your theory seems to address what the peaks should look like, how to identify one, not how they are made or what they are made of."

Yes, that is correct.

"I don't want to distract from discussion of your presentation, but I cannot help but compare it to mine, which would, in this case, be more about how they are made."

I would love to see your own "presentation", please where can I find it ?  As for "how they were made", I am unclear whether you mean what causes the pattern, how the pattern causes life, or how we build machine life with my Theory as a basic template.

"Yours seems more behaviorist, more diagnostic, than analytical."

I can accept that assessment. Although if you don't know exactly how an Agent must behave in order to be alive and intelligent, you would have to be pretty lucky to build one successfully.

"You may be more hardware while I am more software (not that there's anything wrong with that, it's a good thing)."

Actually, I'ld have to say it's the opposite. I work more with software and integration with hardware, and only a little in the building of new hardware.

"I assumed (and infer from your more recent posts) that you use your diagnostic theory to assist you in the more developmental work that you allude to."

Absolutely correct, yes. But again I reiterate, I think that my Theory still opens up huge areas of exploration in metaphysics, etc.

"Truthfully, yours is a codification of what I and many others probably subconsciously assumed, but never got around to verbalizing,..."

Yes, one of the big problems is that some people write that my Theory has already been laid down, because it just seems like it should have been. But when I challenge them to produce anything that resembles my Theory, they can not.

"...and mine, if I get it summarized into intelligibility, will also be seen as what everyone already knows intuitively."

I truly look forward to seeing this !

"The first person to invent the wheel was probably in the same boat, his brother-in-law said, "Yeah, yeah, round things roll... so what?"."

I have faced since I published a pretty much endless stream of people with exactly the attitude you describe. It is frustrating, for sure. I am hoping to set the record straight here in these forums so that MAYBE 10 or 15 (with my luck, it will be 50) years from now I might actually get credit for some insights. In the meantime, I am a tough old goat - I am to stubborn to quit.

"Then suddenly his brother-in-law is making wheels for a living and the Universe is forever changed."

HA !  That was hysterical !  And by the way, you have an awesome and very pleasant writing style. Your work, when you publish - will be not only lucid, but very humorous and uplifting as well.

"You have set a clear method of discovering if that has happened..."

If the Agent in question contains any of my Elements, then it has happened. That's my method.

"...which can be of significant value only if we ever figure out how to make it happen!"

I believe we long ago crossed that threshold, we are easily making "it" happen.

"Asserting a complex enough system gives us so many switches that there are not enough Chronons in the Multiverse to test a statistically significant portion of the possible combinations. Random chance alone requires a blase' invocation of the Large Anthropic Principle to get past.

While I don't require a personified Designer for the Design, I would like to see a mechanism. The Final Anthropic Prinicple asserts that Intelligence must form in any universe like ours, but cannot explain how that might happen."

A great metaphysical exploration. There are a few terms there with which I am unfamiliar, but nevertheless interesting.

"While I don't require a personified Designer for the Design, I would like to see a mechanism. The Final Anthropic Prinicple asserts that Intelligence must form in any universe like ours, but cannot explain how that might happen."

Again, awesome questions to explore. But I must add they are irrelevant to actually building an MLAI Agent. And that is the purpose of my Theory.

"My cat requires me to feed it or it will die."

So ?  If your cat runs away and becomes feral, your point is mute - it will feed itself. But even if this were not so, an Agent does not have to exhibit an Element in it's perfection. In fact even individual humans would not score perfectly on every Element if we benchmarked them. All of the Elements are gradients. Your well fed cat still cleans itself and (hopefully) poops in it's box, etc. And it's body heals itself, and healing is another aspect of this Element.

"Handy, but not required? If it is made to work right to begin with, it will work right...."

Perhaps for the most simple of Agents. But for extremely advanced ones it would be almost impossible to pre-program every single environmental possibility.

"That establishes evolution, not Life. Many things were alive which could not adapt and have now died."

I do not understand the allusion to evolution. But procreation is just one Element of Life, and as I have said - not all the Elements need be present for the Agent to be alive. And finally, being alive and being successful as a species are two different things.

"This restates 1 and 2 combined, doesn't it?"

1 and 2 do not make any mention of components or their relationship to each other. I can not grasp the connection you see between procreation and complexity. I suppose certain adaptions might work towards increasing the Agents complexity - along with many other factors of course. Certainly I disagree that it is a restatement in any way.

"This is our limited idea of what is required to accomplish 1,2,3 and 6. We are probably right, that we need to sense our world to Eat, Grow, Breed and Run for our lives! But there may be another way, such as be created in a universe with a handy human to do all that for us."

This statement is true, though I struggle to see the relevance.

"Assumes dangers exist."

Danger always exists. The universe appears fundamentally designed to tear this pattern to shreds via entropy and chaos. Somewhere someone asked me why I would want to build fight or flight into the skittering little insect bots, etc. that I have running around my place here, well - it's simple. My dog Bethoven HATES them. I have learned a great deal about fight or flight from trying to teach/program these little guys to not get killed by my dog. Have to say it's just a blast to watch, to.

"Agree with "preserve information", but suggest that "organize and utilize" are not necessarily part of memory."

What possible use could a memory be to an Agent if he can't recall the data at a later time ?  And recall would therefore be the "utilization", and that process would in turn require that it be somehow organized to enable this. A tape recorder "preserves" information, but can it recall that information and make use of it to it's own benifit ?

"I find this a harsh limitation of "Self Awareness"."

It's not meant to be. It's meant to be very minimilistic though. The absolute resolution of the Element.

The dictionary defines self awareness simply as knowledge of one self. I actually take it a step further, requiring at least two objects. This is pretty much all self-awareness is. But most people do tend to take it as common knowledge that self-awareness is some big esoteric thing. But like all the Elements, it's a gradient. I have stated the bare minimum necessary to be self-aware, but if you continue to add objects and increase the Agents resolution of those objects - it could become "a higher order of abstraction, where a process can monitor its own process".

"Again, a human limitation. I would allow it based upon the idea that we need at least this level of anthropomorphism to be able to recognize anything, but I would like to see a more universal standard. I think you approach it in your 6)."

You are correct, and have hit upon the only modification I have so far considered based on these critiques. Based upon your own and others thoughts on this, I am tentatively considering adding a single word to my Theory and changing the name of that Element to "Higher Intelligence". And that is because I think that lower intelligence, or whats often referred to as "animal intelligence" is inferred by the other Elements - which elucidate many behaviors that could be deemed intelligent.

"Show me an average human!8-)"

CLICK  HERE

"This is the whole problem, we have to use our standards to measure "The Other".... a basic fallacy."

I agree, but until we run across some other sentient races, it's the best we can do.

"Here we come to Self Awareness. For an Agent to choose a new state, it must, on some level, apprehend its current state."

I agree with this statement.

"And here is where the rest of your "Memory" above might belong. Because unless you are talking about reflex and instinct, which I think you are not, organization and utility come after reduction."

I agree that a properly working memory would be required for the reductions to be stored and utilized. But that would be just one thing the Agent would be using it's memory for. Remembering where it's food (or electricity plug) is located, remembering how to run away from the huge slobbering hairy monster that wants to chew your head off, etc.

"I think you need to define a threshold level. Perhaps values could be assigned which are additive, allowing scoring on a numerical basis."

I have done all of these things. I have devised a graded testing regimen for all of the Elements. I have been using it successfully in my own work now for almost a year. But the Theory purposefully does not draw a line in the sand, and allows for the fact that we can not currently resolve the boundary between life and not-life by leaving it up to the tester to set the parameters. I am releasing the Theory first, because the testing regimen is irrelevant to the logical defense of the Theory itself, and I don't want to get bogged down in defending them both at the same time (this is hard work !  I released this in 10 of the biggest MLAI forums and I am now spending 10 to 12 hours a day writing these posts !). Once this process dies down a little, I will tidy up my lab notes and release the testing regimen and criteria I am using. But anyone can devise simple tests to do as you state above. Several such tests are mentioned in the works of Turing, Von Neumann, and Wagman. First the Theory, then it's defense and explanation, then it's application. I am to old to do it all at once !

"IN the threshold spectrum, we need to realize that we are only in the visible rainbow range. Perhaps there are levels under us that we can't see, and levels above."

He, he - okay.   :rolleyes

"Our ability to measure requires that our definitions only include that which we can measure. Perfect circle of logic."

A central pillar upon which my Theory is based.


YOURS -- Christopher Doyon    :cheesy



MLAI  Foundation

www.MLAIFoundation.info

*

Art

  • At the end of the game, the King and Pawn go into the same box.
  • Trusty Member
  • **********************
  • Colossus
  • *
  • 5864
Re: A Unified Theory Of Machine Life And Artificial Intelligence
« Reply #14 on: November 17, 2006, 10:14:37 am »
Chris,

Is your bot project an original, stand-alone creation capable of learning OR
is it a super scripted Alice bot?

In the world of AI, it's the thought that counts!

 


Useful new robot.
by MagnusWootton (Robotics News)
January 15, 2022, 01:50:10 am
Christmas Eve Launch
by frankinstien (Robotics News)
January 13, 2022, 08:08:13 am
Forest - The Show
by Zero (Robotics News)
January 09, 2022, 01:58:03 pm
I wonder how well GPT-3 will on run this.
by frankinstien (AI News )
December 20, 2021, 04:36:39 am

Users Online

44 Guests, 2 Users
Users active in past 15 minutes:
chattable, ivan.moony
[Nomad]
[Trusty Member]

Most Online Today: 64. Most Online Ever: 2369 (November 21, 2020, 04:08:13 pm)

Articles