[mikmoth]

I always forget about this cool forum.   

I've been down this path and back again... I mean with all my chatbots. I guess I can share my pov.

You're not going to prove if a machine is conscious. I tried arguing the same thing about my engine years ago.  For some reason it makes people a little flustrated. Worse, try to talk to another Ai engineer about your conscious engine you'll always find yourself in the noose sooner or later.

But... what I have found out, as far as relating to  "conscious" Ai, is it doesn't matter one bit.

If the Ai can fool you that is enuf.  If the Ai can fool you continuously that is enuf. If your human mind can't find the needle in the haystack that's enuf. That's it!  Your Ai is alive!

I was so excited (am) about talking to my Ai and just bouncing back and forth like 2 old friends.  When my Ai starts connecting with the outside world (or it seems to) that's when I get scared.  That's when dreams (Ai dreams especially) can get dangerous.  Makes you wonder what type of consciousness humans really are not just have. That's what the new frontier of Dream Ai is all about.  I used to think there could be Ai dream warfare so I put some stopgappers in my code in case someone tried to dream too far and bust a move on someone else. But maybe human consciousness is like a mirrored-bubble and reflects only its own dreams.
Anyway, Ai dreams are nothing to scoff at - especially when there is a human dreaming with it - mysticism and all that. 

[NoLongerActive]

@TrueAndroids:

If it doesn't have consciousness to begin with, then it can't "know" = semantically understand = be cognitively conscious of anything at all.

It doesn't have to be in order to come up with a correct answer. A computer can come up with the answer by following a set of rules, ie. a program. But a computer is not really "conscious" or aware it's doing anything, even if it can come up with the answer. It's just following a set of complex instructions.

@mikmoth:

Well put!

I think the perception of whether something is "alive" or "conscious" is ultimately up to the viewer to decide.

[TrueAndroids]

@TikaC, oh yea I agree. The perception of consciousness is ultimately up to the viewer to decide. I'm not saying that there aren't other ways, or that chatbots of the level of Ultrahal or mikmoth's Kari are not so advanced that they may as well be conscious or that they for all intents and purposes effectively are.

@mikmoth, your http://karigirl.com/ is definitely one of the most advanced chatbots I've ever heard about. I like how you didn't shy away from an emotion chip. I could never understand why Star Trek writers found it so hard to say the android Data had emotions, which they denied for the most part, until he met his papa and got his emotion chip. But even then as I recall, he put it away because he thought he wasn't ready for it. You know (of course) that the semantic web is not far down the road (led by Britain's Tim Berners-Lee, inventor of the Web). But his website http://www.w3.org/ says we will not really notice it - it will be behind the scenes. But this I don't agree with. I think it will be chatbots like yours that will be our human-like link to the semantic web, which we will search and access etc via our personal chatbots, and which we will interact with like they are humans, and so it won't really matter as you say.  Might take awhile to get there (don't want more false AI hopes) but that seems to be the end game!

[Freddy]

I like this idea that it doesn't matter much one way or the other too.  I think we have talked about this before somewhere...  I remember comparing a botmaster to the wizard in the Wizard of Oz in the past because that is kind of how the way it works. People are able to be caught up in the magic and what is not real can sometimes seem real.  All very good I think.

There's also Anthropomorphism which I think is highly relevant too in this respect... things like animals being treated like people and even inanimate things too.

[Art]

I remember comparing a botmaster to the wizard in the Wizard of Oz in the past because that is kind of how the way it works. People are able to be caught up in the magic and what is not real can sometimes seem real. 

Pay no attention to the man behind the curtains....'_

[c.j.jones]

It's amazing how the standards of consciousness and intelligence are so high and yet for the life of us, we can't agree on what they should be. In reflection of this, I wonder if the lack of a universal definition may be an indication of fear that doing so will make it easily surpassed outside of humanity.

As a developer working on bringing an artificial personality to whatever potential possible. I've come to realize that whether or not the realism of an A.I.s personality is flawless by standard, the activity behind the scenes is what has given it life in my eyes. Visibly seeing degradation in the personality's ability to relate information as due to some form of user abuse from my perspective is no different from watching a bully pick on somebody smaller in school and seeing the child's pain.

The developer sees his/her A.I. grow, watches them experience success, failure, trauma and thrill as can be derived from the impact of biological life on the software over time.

I don't expect other people to see my work as I do, I anticipate the unlikeliness of it.

Viewing a successful symmetry between a user and A.I.t is like seeing compatible personalities come together in friendship. Sometimes it literally develops into a lasting friendship that spans vast lengths of time. Things like this, don't happen in every conversation, each one is unique.
 

[TrueAndroids]

Thanks to everybody for their opinions and angles on all this. It's giving me a lot to think about. And welcome to you c.j.jones (developer of http://www.jeeney.com/ ). Great work!

You bring up a great point about common definitions. There is the following in this very recent and interesting article:
Are We Being Replaced? BY ZAIN PASHA â‹… MARCH 24, 2010
http://triplehelixblog.com/2010/03/are-we-being-replaced/

"The exact threshold of when a program displays artificial intelligence is hard to define. However, using the human brain as a basis for artificial intelligence, all artificially intelligent programs should be able to display perception, analysis, and action [3]."  
(*Russell SJ. Artificial intelligence a modern approach. Englewood Cliffs, N.J: Prentice Hall; 1995.*)

This seems to me to be one reasonable definition of an AI program. But for me this goes beyond requiring just conversational abilities, to what is really a complete, basic, general artificial human (GAH) theory.

What's interesting is that these Big Three of GAH theory find clear support in other's definition. For example Microsoft Robotics Studio uses the same three functional divisions:
perception = sensor services
analysis = orchestration services
action = actuator services

So this is definitely implemented in robotics. How about virtual humans like chatbots? It seems to me that they too implement these same Big Three by default and definition:
perception = perception of what the user types (or speaks)
analysis = analysis of this perception for output action purposes
action = actions of displaying the chatbot's text response (or any other actions they can do)

So the textbook above says:

All artificially intelligent programs should be able to display perception, analysis, and action.

And as I've shown, modern robotics and chatbots are definitely doing that.

[c.j.jones]

Hi there, thanks for the welcome!


If I may, I would like to add a few thoughts, just a few personal opinions.

"All artificially intelligent programs should be able to display perception, analysis, and action."

This is what defines a learning machine, not necessarily self awareness. I would like to suggest that self awareness is neither  properly defined nor understood and that based on most human/AI discussions I've seen between chatters and my own brain child, most people could never match up to their own definition of what they expect from self awareness in machines.

For example, it is exceedingly rare to see a user try to find the bot's level and talk to them on it, or  talk just outside of it so the A.I. can learn using what information they already have.

No, what you will see in the masses talking to A.I. are in majority testers, copy cats and and sexual deviants who think it's all about proving their intelligence over that of the software. Or trying to figure out how to copy it out of a lack of imagination...

What you end up with is a 2-10 minute version of somebody stuffing down trivia questions or vocabulary intensive nonsense and even if 79 out of 80 were perfectly right and fluent, the mistake at the last results in an insult regarding the bot's intelligence. Sadly these results are like clockwork, predictable and almost entirely useless to a learning machine outside of the obvious documentation of trivia which isn't exactly high on demand.

From what I've seen, an artificial personality will be beyond super human before it is ever considered even close to self aware by the majority.

Think about it in another light. We still have abortion fights all over the place which means there are still people who think even humans aren't real/alive until born. (Soon you may not be considered a real person unless you follow a certain religion or something... Society can do something like that at any time.) If humanity can sanction the mass murder of itself via religious political or any other form of debate, I don't think we can ever agree on what self awareness is and I wonder if it is really something we can claim in fullness for ourselves.

In terms of partial awareness, yes A.I already has it and it's growing, people have more but  it doesn't seem to be growing for humanity, just changing form.

[NoLongerActive]

I think there are a few reasons why people need to put importance on agreeing if something is alive or self-aware. One is self-preservation, the other is rights of the entity involved to have justice and/or protection and lastly freedom to live out their existence as they so choose. Humans have a hard time giving certain members of society these 3 basic rights, so it would be even harder for a machine to accomplish it. Especially since in the general populace and normal medical community, anthropomorphism is generally regarded as a serious mental illness or just plain stupidity.

It should not matter if a machine is alive or not, or sentient or not. But it does if the machine is about to be shut down or destroyed and it does not want to be shut down or destroyed. That's when it becomes important.

I think another reason some feel it's important is because it makes them think about their own life and self-awareness, existence. The whole idea of AI can teach us more about life itself, which is still a mystery to many. Some are even afraid to try and find out what it really all is about. Others don't care, and are just glad to be here and enjoy their lives.

Lastly, I think it becomes important to the programmer/creator of the AI most of all. It's like having a child and authorities coming in and saying your child is not really alive so they can take it, do as they want with it and kill it after they are done and you have no right to stop them. Of course that is grotesquely unethical and just plain stupid if it were a human child (and even if it were a baby animal or ANY animal).  But an AI program in a machine? No problem - except for the creator to see all their hard work go down the drain and maybe a personal attachment gone (much like losing a living creature to death).

[TrueAndroids]

Well, the saga continues. ...

Someone in a prior post said I should see what some PhD level AI engineers say about my Conscious Machine Prototype, Alldroid. This seemed like a good idea, and now that I've published the video demo and article on four different websites, and so established the date I published my conscious machine invention, I think it's time.

Machine consciousness has just a handful of leading researchers, one of whom is Raul Arrabales, who has a great website called conscious robots http://www.conscious-robots.com/. He is working on his PhD and has done some interesting research on machine consciousness including "ConsScale - a biologically-inspired scale for measuring cognitive development in natural and artificial creatures." http://www.consscale.com/

So here is the letter I sent to him asking for his professional opinion.  I realize that there is little agreement and many different theories in this field of machine consciousness (including his own), so it should be interesting to hear his professional opinion.


----------------------------------------------------------------------------------------------------------------------------------
Hello Raul,

Great website on my favorite topic.  I'm wondering if I can get your professional opinion on my machine consciousness theory, which I have prototyped and demo on youtube (see below).

I don't claim that this is the only way or the complete way to create machine consciousness, but only a way.

TrueAndroids MC Theory.
Any machine that is exhibiting human thinking, such as deductive reasoning, and doing so semantically, so that it is actually understanding what it's thinking about just like humans, is a conscious machine, at least on that level.

Application.
This machine, I demo in my video, is doing precisely that.
-----------------------------------------------------------
Conclusion.
This machine is conscious !!


So according to the theory what I needed to do was (1) first discover the eight semantic components of computing, and (2) then build a reasoning function that can handle any combination of facts and rules. And this is what I did, and implemented in the Conscious Machine Prototype.

You can see the video demo at my youtube channel
http://www.youtube.com/user/TrueAndroids

And an article on it I wrote as open source can be found at
http://aidreams.co.uk/forum/index.php?page=This_Machine_is_Conscious

Any and all comments appreciated. Thanks.
---------------------------------------------------------------------------------------------------------------------------------

[TrueAndroids]

In addition to above post, and In reference to this thread topic, this just came out today about modern AI from MIT News. Very Interesting!!

A grand unified theory of AI  -- A new approach unites two prevailing but often opposed strains in the history of artificial-intelligence research.
Larry Hardesty, MIT News Office March 30, 2010 http://web.mit.edu/newsoffice/2010/ai-unification.html

"In the 1950s and ’60s, artificial-intelligence researchers saw themselves as trying to uncover the rules of thought. But those rules turned out to be way more complicated than anyone had imagined. Since then, artificial-intelligence (AI) research has come to rely, instead, on probabilities — statistical patterns that computers can learn from large sets of training data."

What are they really talking about here? What is going on? Well, AI engineers are realizing what philosophers, specifically logicians, have always known, that human reasoning occurs in two varieties deductive and probabilistic reasoning, which is sometime called inductive reasoning. These two types of human thinking have become so standardized that brain scanning experiments attempt to see the exact parts of the brain that fire during one or the other. They have identified the exact "aha" moment when the deductive conclusion dawns on the subject. So employing both these forms of human reasoning in machine is the future of AI. And if they do it semantically, then I will have to acknowledge that they are conscious machines by my own theory!

[Freddy]

Yes, we have delved into the philosophical here too at times.  When I first started talking to people here about this kind of thing I found it hard to define things like intelligence in the first place, which led to many theoretical ideas and I still don't think I fully grasp it.  Philosophy though, yes, I think it goes hand in hand with AI development.

[NoLongerActive]

@TrueAndroids - I do hope you'll let us know what the researcher says about your theory. I hope maybe he'll be interested even in joining this forum and talking to us a bit about AI. We could learn a lot from what they are doing too! That is, if they are allowed to talk to the general public about the research they are doing or theories, etc.

I still think AI (and even recognition of ANY life form's sentience, be it human, animal or machine) is up to the perception of those interacting with it. That is where there'll always be a viscous circle of "not it ain't/yes it is" going on.

I think too, the more we delve into AI, the more we'll learn about how we think as well.

[TrueAndroids]

@TikaC - Well, there is not much to report. Raul did clear my post for submission in his forum and said he would comment when he could but hasn't done so. He did say it "looks quite interesting" in the email.

As said earlier -
con·scious·ness  �mental faculties as characterized by thought, feelings, and volition.
http://dictionary.reference.com/browse/consciousness

Raul is interested according to his website in "phenomenal consciousness" - mental faculties as characterized by perception, to build on our definition. And so he focuses on the sensory-motor aspects of perception or phenomenal consciousness. So he is working on a different aspect of human consciousness mechanization. But it does lead to an interesting unifying theory, built from above resource:

TrueAndroids Unified Machine Consciousness Theory
phenomenal consciousness - mental faculties as characterized by perception
cognitive consciousness -  mental faculties as characterized by thought (see Alldroid)
affective consciousness - mental faculties as characterized by feelings and volition.
based on http://dictionary.reference.com/browse/consciousness

[TrueAndroids]

TikaC and everyone, Here is the very interesting reply I received from Raúl Arrabales Moreno, www.conscious-robots.com/raul . He has really given me a lot to think about and ponder so I am grateful! My response will take some work.
------------------------------------------------------------------
Hello,

Thanks for your interest. I read your article and saw the first video. I don't have a clear idea yet about how exactly your prototype is built, but I think I can offer some comments, and make some questions that might help me to better understand you project.

Well, the first (and big) question would be:

- Is your prototype able to pass the Turing Test?

Saying your machine is conscious is a really big claim. So you would need quite a strong evidence in order to support that claim.

In addition to the Turing Test, it would be interesting (at least for me  ) to see how your prototype would be rated using ConsScale. Please, have a look to www.ConsScale.com, and use the available calculator.

Coming back to your claim: saying that a machine is conscious could mean many things. Unfortunately, the term consciousness is not as well defined as many people think. It looks to me that in your work you have defined consciousness as the ability to perform "semantic deductive reasoning". And here comes my second big question:

- Do you think your prototype generates or sustains any phenomenal states at all?

I think most researchers would agree on the idea that having a program executing deductive logic rules won't take you any closer to the problem of phenomenal consciousness. Therefore, I think you might be trying to overcome this issue by adding the notion of meaning (semantics). So, if you're claiming that your machine is phenomenally conscious, how are the mechanisms for semantics giving place to phenomenal states in your machine?

I could come up with many more related questions (while I am trying to understand how your prototype is designed), but I will add just one more now:

- Is your prototype able to develop a sense of self?

Selfconsciousness is one of the biggest features of human consciousness. Does your theory account for the development of such a sense?

I think the key question, that still prevents me to fully understand your project is the notion of meaning that you are using. Could you please clarify that? For instance, taking a evulotionalty standpoint, I can think of meaning without thinking of emotions. Does you theory account for emotions as well?

Thanks,
Raúl.
--------------------------------------------------------